1. business and industrial
2. chemicals industry
3. dyes and pigments

Journal of Fluorescence, Vol. 15, No. 3, May 2005 (© 2005)
DOI: 10.1007/s10895-005-2630-3
How to Improve Quality Assurance in Fluorometry:
Fluorescence-Inherent Sources of Error and Suited
Fluorescence Standards
U. Resch-Genger,1,6 K. Hoffmann,1 W. Nietfeld,2 A. Engel,3 J. Neukammer,4
R. Nitschke,5 B. Ebert,4 and R. Macdonald4
Received October 15, 2004; accepted January 28, 2005
The scope of this paper is to illustrate the need for an improved quality assurance in fluorometry.
For this purpose, instrumental sources of error and their influences on the reliability and comparability of fluorescence data are highlighted for frequently used photoluminescence techniques ranging
from conventional macro- and microfluorometry over fluorescence microscopy and flow cytometry
to microarray technology as well as in vivo fluorescence imaging. Particularly, the need for and
requirements on fluorescence standards for the characterization and performance validation of fluorescence instruments, to enhance the comparability of fluorescence data, and to enable quantitative
fluorescence analysis are discussed. Special emphasis is dedicated to spectral fluorescence standards
and fluorescence intensity standards.
KEY WORDS: Fluorescence; standard; calibration; microarray; in vivo imaging; flow cytometry.
non- or minimally invasive character, and the remote accessibility of signals employing conventional optics and
fiber optics. Depending on the specific application, fluorescence measuring systems strongly differ in instrumental design, i.e. use of optical components and detection systems as well as in measurement geometries
including sample containers/formats for the investigation
of fluorescent samples. Typical instruments range from
conventional monochromator- or filter-type fluorometers
and microplate readers over fluorescence microscopes,
flow cytometers, and systems for (real time) polymerase
chain reaction (rt-PCR) as well as devices for molecular
imaging to microarray scanners for DNA or protein chips
and high throughput screening systems [5,7–10]. Independent of the technique and particular instrumentation
applied, a major drawback of fluorescence measurements
is the demand to remove instrument-specific effects from
the measured raw data that otherwise limit the comparability of fluorescence data across instruments, laboratories, and over time [1–4,11,12]. Further fluorescenceinherent problems are related to difficulties to accurately
measure absolute fluorescence intensities and hence to
INTRODUCTION
Fluorescence techniques are widely used as analytical tools and detection methods in different areas such
as material sciences, environmental analysis, bioanalysis, molecular genetics, cell biology, medical diagnostics, and drug screening [1–5]. Commonly recognized
advantages of fluorescence are its high sensitivity that
allows even the detection of single molecules, its intrinsic selectivity concerning experimental parameters such
as excitation and emission wavelength as well as fluorescence lifetime and (de)polarization [6], its ease of use,
1
2
3
4
5
6
Federal Institute for Materials Research and Testing, Working Group
Optical Spectroscopy, Division I.3, BAM, Berlin, Germany.
Max Planck Institute for Molecular Genetics, Berlin, Germany.
Schott AG, Mainz, Germany.
Physikalisch-Technische Bundesanstalt, Dept. 8.3, PTB, Berlin,
Germany.
Life Imaging Center, Institute of Biology I, Developmental Biology,
Freiburg, Germany.
To whom correspondence should be addressed. E-mail: ute.resch@
bam.de
337
C 2005 Springer Science+Business Media, Inc.
1053-0509/05/0500-0337/0 338
realize true quantitative measurements [13]. In addition,
quantitative fluorometry based on measurements of relative fluorescence intensities is hampered by the sensitivity of the absorption and fluorescence properties of most
chromophores to their microenvironment that results in
environment-specific absorption and emission spectra as
well as molar absorption coefficients, fluorescence quantum yields, and fluorescence lifetimes, respectively. These
problems and disadvantages render quality control difficult and make standardization even more important.
The improvement in quality assurance for fluorescence techniques critically relies on the development of
internationally accepted and purpose-fit guidelines for
instrument characterization and performance of fluorescence measurements which are barely available at present.
The former, however, depends on the availability of commercial, easy-to-operate, and reliable fluorescence standards, that take special requirements of each particular fluorescence technique into account, and that are preferably
certified in view of traceability and accreditation needs.
Within the context of this paper, fluorescence standards
refer to chromophore-based reference materials for fluorometric quantities, which are designed for specific fluorescence applications. This includes both instrument-type
standards for the calibration and performance validation
of fluorescence instruments as well as more applicationspecific standards for instance to relate chemical concentration to instrument response, i.e. typically to fluorescence intensity. Furthermore, for certain fluorescence
techniques, the evaluation of commonly used software
routines like e.g. fitting routines for the analysis of timeresolved fluorescence data as well as software for the
statistical analysis of imaging data and the analysis of
microarray experiments is gaining importance and will
eventually be mandatory.
At present, these needs are poorly met for fluorescence techniques, despite of their widespread use and the
extensive literature dedicated to the characterization of
fluorescence measuring equipment and potential fluorescence standards [1–4,14–18], as well as the globalizationinduced anchor role of traceability and accreditation [19].
The latter initiated for instance standardization in closely
related fields like UV/Vis/IR spectrophotometry [20–23]
and colorimetry [24–27]. Up to now, there exist only a very
limited number of recommendations on the characterization of fluorescence instrumentation and performance of
fluorescence measurements like the determination of the
linear range of detection systems and the limit of detection [28–30]. Furthermore, with the exception of flow
cytometry [31–34], surveys on instrument calibration and
comparability of fluorescence data have been barely performed [35,36]. In addition, available or recommended
Resch-Genger et al.
fluorescence standards are often of insufficient quality
with respect to their analytical and spectroscopic characterization and stability as well as reproducibility or are
even not purpose-fit. For a long time, only a single certified
reference material [37], the traceable emission standard
quinine sulfate dihydrate (SRM 936a), developed by the
National Institute of Standards and Technology (NIST) in
the late 1970s, was commercially available [38]. This reference material enables the determination of the spectral
responsivity of fluorescence instruments in the spectral
range from about 395 to 565 nm. Only recently, a new
though not certified reference material, RM8640, i.e. microspheres with varying amounts of immobilized fluorescein molecules was released by NIST for flow cytometry
thereby aiming at the establishment of a reference scale
for fluorescence intensity [39].
To underline the importance of an improved quality control for the majority of fluorescence techniques, in
this review, problems and fluorescence-inherent sources
of error are revealed exemplary for macro- and microfluorometry, fluorescence microscopy, flow cytometry, fluorescence-based microarray technology, and
in vivo fluorescence spectroscopy/imaging. The aim is
to highlight the need for internationally accepted guidelines for instrument characterization and performance of
typical fluorescence measurements in combination with
the need for better-suited and certified fluorescence standards thereby improving the reliability of fluorescence
data. This could eventually broaden the application of
fluorescence methods, especially in areas like medical diagnostics that require instrument performance validation
(IPV), standardization of measurements, and comparability of data for the acceptance of analytical methods. Accordingly, the scope of this article is to discuss purpose-fit
solutions to these problems with special emphasis dedicated to specific existing fluorescence standards and the
need for and requirements on improved systems rather
then to give a comprehensive overview of the literature
on the characterization of fluorescence instruments and
potential reference materials.
FLUORESCENCE-INHERENT SOURCES
OF SYSTEMATIC ERRORS
The majority of fluorescence-based measuring or detection methods relies on the assumption that the recorded
fluorescence intensity is proportional to the concentration of the fluorophore(s) in the sample. However, independent of fluorescence technique and type of measurement, i.e. spectrally resolved or integrated over a small
wavelength interval, the generated fluorescence signal is
Quality Assurance in Fluorometry and Suited Fluorescence Standards
affected by both the fluorescent analyte(s) and the instrument. Properties of the sample that influence the magnitude of the resulting fluorescence signal are the fluorophore’s absorptance at the excitation wavelength (interval) and the fluorescence quantum yield [40,41]. The
former is nonlinearly linked to absorbance and thus molar absorption coefficient and concentration by the Beer–
Lambert law. Absorptance and quantum yield as well as
the spectral shape and position of a chromophore’s absorption, excitation and emission spectrum can strongly
depend on the dye’s microenvironment, e.g. the polarity
of the surrounding solvent/matrix molecules, pH as well
as the concentration of certain ions in the sample [1–5,13].
Furthermore, specific solvent–solute or matrix–solute interactions can play a role. From the instrument side, the
time-, wavelength-, and polarization-dependent spectral
irradiance of the excitation channel reaching the sample
and the spectral responsivity of the emission channel influence the magnitude of the recorded fluorescence intensity and the shape of the measured fluorescence spectra,
respectively [40,42]. To illustrate the distorting effects of
the spectral characteristics of fluorescence measuring systems, the normalized spectral responsivities of common
types of fluorescence instruments are exemplary shown in
Fig. 1. These curves have been determined with a spectral radiance transfer standard [40]. Accordingly, aside
from spectral irradiance-related effects, measured emission intensities or spectra represent truly sample-related
data that are superimposed by the spectral responsivity of
Fig. 1. Wavelength-dependent normalized spectral responsivity of the
emission channel of four typical fluorometers equaling the corresponding normalized emission correction curves. Spectrally corrected emission spectra are obtained on division of the measured uncorrected spectra
by the respective emission correction curve measured with identical instrument parameters.
339
the respective instrument, with the size of these distortions
depending on the spectral region of the dye’s emission and
the width of its emission band. Similarly, as has been illustrated in Parts I and II of a series of publications dedicated
to the traceability in fluorometry [40,42], the instrumentspecific wavelength and polarization dependence of the
spectral irradiance at the sample position distorts excitation spectra and affects the intensity of emission spectra.
Other potential sources of error are uncertainties of
the wavelength scale which need to be minimized especially for fluorophores with narrow absorption and/or fluorescence bands and nonlinearities of the detection system.
The latter can hamper quantitative fluorometry and the determination of fluorescence quantum yields [40]. Furthermore, instrument- and sample-related polarization effects
may play a role [39,43]. The size of instrument-related polarization effects is determined by the dependence of the
transmittance and reflectance of the instrument’s optical
components on the polarization of the incident light and, to
a much smaller extent, by polarization effects of detectors
[40]. The magnitude of sample-related polarization effects
depends on the fluorescence anisotropy of the sample.
Instrument- and sample-inherent polarization effects, that
are strongest for the measurement of anisotropic emitters
without polarizers, can yield intensity and spectral errors
in fluorescence measurements in the range of a few tens
of percent [15]. Depending on the fluorescence technique,
additional uncertainties can be introduced by e.g. an insufficient long-term instrument stability or instrument drift,
an inhomogeneous illumination of the sample as well as
generally by sample- and sample-handling-related effects.
The latter are beyond the scope of this article.
As is often overlooked for certain analytes/samples,
also type of light source, i.e. continuous or pulsed excitation, and related instrument settings such as e.g. the width
of the time gate and delay have to be considered. This is
illustrated in Fig. 2 for the normalized corrected emission
spectra of a glass doped with rare-earth (RE) ions displaying different lifetimes in the µs- and ms-range that
was measured with conventional steady-state fluorometers equipped with a continuous and a pulsed Xe lamp
(width at half peak intensity of less than 10 µs), respectively. Such a RE-doped glass is an interesting candidate
for a standard to control the wavelength accuracy and
long-term stability of fluorescence instruments [44]. The
differences observed in Fig. 2 are caused by the different and long lifetimes of the dopants and are not related
to spectral correction. Basically, with the settings employed for pulsed excitation, suppression of the long-lived
emission compared to the emission of the species with
shorter lifetimes occurs. This observation correlates with
the width of the time gate used for signal detection. Only
340
Fig. 2. Comparison of the corrected emission spectra of a glass, doped
with a mixture of rare-earth (RE) ions, measured with a fluorometer
equipped with a continuous excitation source (8100, Spectronics Instruments, Xe lamp, solid line) and a fluorometer with a pulsed lamp (LS
50B, Perkin Elmer, width at half peak intensity of the pulsed Xe lamp
less than 10 µs, fluorescence mode, dashed line) using identical spectral
bandpasses of the emission monochromators. Excitation was at 370 nm.
minor effects are due to subtraction of dark current contributions and background from the long-lived emission.
Accordingly, for instruments equipped with pulsed light
sources, these experimental settings have to be considered
for the comparability and reliability of fluorescence data
if species with sufficiently long and different lifetimes are
analyzed simultaneously. Furthermore, these parameters
need to be kept constant for the control of the day-to-day
performance and long-term stability of such fluorescence
instruments.
Aside from instrument-specific effects, the reliability of fluorescence measurements as well as the limit of
detection can be affected by non-specific or so-called
background fluorescence. This includes all the fluorescence that does not originate from the fluorophore of
interest. Background fluorescence may arise from contaminations of the sample and non-specific binding as
well as from autofluorescence of optical components such
as lenses, objectives, and optical filters [45] and sample supports/containers like glass or polymer slides and
microtiter plates, respectively. The possible influence of
autofluorescence from optical components is illustrated
exemplary for filter glass OG 550 obtained by different
pre-treatment procedures, see Fig. 3. Due to its clearly
measurable autofluorescence on excitation at 450 nm, the
emissive filter is not suited for the fabrication of fluorescence instruments working in the UV/Vis region. These
findings underline the importance of an improved quality
control of such materials.
Resch-Genger et al.
Fig. 3. Uncorrected emission spectra of two OG 550 filters obtained by
different pre-treatment procedures recorded under identical conditions.
Excitation was at 450 nm.
FLUORESCENCE STANDARDS: TYPES AND
GENERAL REQUIREMENTS
Types of Fluorescence Standards
The comparability of fluorescence data between instruments, laboratories, and over time as well as the minimization of fluorescence measurement-inherent sources
of errors require a reliable instrument characterization
and performance validation. This can be achieved with
accordingly designed, well-characterized, and easy-touse fluorescence standards including standard operation
procedures (SOPs) for their use. Additionally, as many
fluorescence experiments are performed for quantification purposes, suited standards, recommendations for their
choice and application as well as SOPs for the measurement of (relative) fluorescence intensities are desired.
Fluorescence standards are typically divided into
instrument-type standards, which allow the determination of an adequate and application-relevant instrument
performance thus ruling out instrumentation as a major
source of variability, and more application-specific standards, which take into account the fluorescence properties of routinely measured samples. The latter are also
referred to as reference standards for instance in fluorescence microscopy. Typical instrument standards are
spectral standards with known, preferably certified corrected, i.e. instrument-independent fluorescence spectra.
Spectral standards can be divided into wavelength standards to control the wavelength accuracy of wavelengthselecting optical components and so-called emission
and excitation standards [14] for the determination of
Quality Assurance in Fluorometry and Suited Fluorescence Standards
the spectral characteristics of fluorescence instruments
[40,46–48]. Examples for more application-specific types
of standards are fluorescence lifetime and fluorescence
(de)polarization standards [16,49], that are beyond the
scope of this article, and the class of fluorescence intensity standards. The latter include fluorescence quantum
yield standards and reference materials to relate chemical concentration to instrument response or, less common, to characterize the autofluorescence of materials
and reagents used for fluorescence analysis. Other equally
requested standards, that belong to the classes of fluorescence intensity and spectral standards, are materials
to control the day-to-day performance and the long-term
stability of fluorescence instruments. At present, there is a
broad variety of more or less suited fluorescence standards
commercially available such as for instance fluorescent
microspheres as alignment standards for fluorescence microscopy and flow cytometry, solid and liquid standards
for the spectral correction of cuvette-type measurements
and microplate readers as well as standards for front surface fluorescence measurements, and solid and liquid standards for intensity calibrations and for the determination
of the linearity of microscopes, flow cytometers, and microplate readers, respectively [42,50]. In addition, there
exists extensive literature on potential fluorescence standards. This includes for example emission and excitation
standards and standards for fluorometric quantities like
fluorescence quantum yield, fluorescence lifetime as well
as fluorescence (de)polarization [1–4,14–15,51,52].
General Requirements on Fluorescence Standards
Generally, the requirements on fluorescence standards depend on the level of uncertainty desired for
fluorescence measurements and accordingly, for instrument characterization. Standard-related contributions to
the overall calibration and measurement uncertainty are
directly linked to the reliability and suitability of the
standard(s) and to the uncertainty of the standard’s certified/reported radiometric/fluorometric quantities that are
relevant for the respective application. For many applications, also the resemblance between the standard
and typically measured samples affect the size of the
standard-related contributions to the overall measurement
uncertainty [40].
A purpose-fit fluorescence standard should enable
the instrument characterization and IPV under analytically
relevant routine conditions, e.g. for similar settings of instrument parameters (detector voltage, spectral bandpass,
polarizer settings, signal intensities/counting rates, magnification etc.) and measurement geometry—preferably
with a minimum and known uncertainty. This implies es-
341
pecially comparable emission characteristics and spectral
radiances or emission intensities of the standard and the
sample, typically within a similar spectral region. The
standard needs to be sufficiently well characterized with
respect to its application-relevant spectroscopic properties. Its calibration-relevant radiometric/fluorescence features should have been determined with a reliably characterized reference instrument. To judge the reliability
of the standard, the calibration procedure and calibration uncertainty of this reference instrument as well as
the chosen instrument settings and measurement geometry need to be provided. Moreover, this eventually enables traceability to the spectral radiance scale and/or
the spectral responsivity scale realized by the black body
radiator and the cryogenic radiometer [40,42] and the
mandatory report of the (wavelength-dependent) uncertainty of the respective fluorometric quantity of the standard. In addition, a specification of the standard’s thermal
and photochemical stability, the temperature dependence
and—for some types of standards also the concentration
dependence—of the reported/certified fluorometric quantity as well as information on storage conditions, shelf
life, scope, and limitations for use are necessary [40].
For physical transfer standards, also recalibration intervals should be given. For chromophore-based reference
materials, the purity and homogeneity of the chromophore
and the solvent or matrix should be stated as well as the
homogeneity of the dye distribution for solid systems.
For particle-type standards, also the particle size distribution should be provided as well as the coloring procedure
like for example staining, i.e. chromophore mainly inside
the particle, or attachment to the particle surface, i.e. chromophore mainly at the particle surface and thus accessible
to solvent.
Type-Specific Requirements
For each application, the choice of a fluorescence
standard requires a thorough understanding of its purpose
and, most important, the requirements it has to fulfill.
A suited wavelength standard must emit a multitude of
very narrow emission bands in the spectral range of interest, commonly the UV/Vis/NIR region, and must be
measurable at sample position employing the typical measurement geometry. Furthermore, the spectral position of
the emission lines/bands should preferably show no or at
least a reported temperature dependence within the commonly encountered room temperature region. These demands are best met by a physical-type transfer standard,
an atomic discharge lamp, ideally of cuvette-shape, typically in combination with a non-fluorescent reflectance or
white standard [40]. Such lamps can be used for all types
342
of luminescence instruments and are not restricted to photoluminescence measuring systems like chemical-type
fluorescence standards. For reduced requirements on spectral resolution, also chromophore-based wavelength standards can be employed. Their advantages are the typically
smaller spectral radiances and similar emission characteristics as encountered for the majority of samples. An example is a commercially available dysprosium-activated
yttrium aluminum garnet Y3−x Dyx Al5 O12 (DYAG, Photon Technology International Inc., FA-2036) that shows
a multitude of narrow excitation bands in the spectral
region between 250 and 500 nm and narrow emission
bands in the spectral region from 500 to 800 nm, respectively [53]. Similarly suited are mixtures of RE ions, see
Fig. 2. Here, however, excitation wavelength-dependent
emission spectra are encountered and accordingly, provision of the emission spectra for frequently used excitation
wavelengths may be appropriate. Contrary to wavelength
standards, the main requirement on emission and excitation standards that enable the determination of the relative spectral responsivity of the emission channel and
the relative spectral irradiance at the sample position, respectively, are broad and unstructured fluorescence spectra to reduce the effect of instrumental resolution/spectral
bandpass [40,47]. This minimizes the influence of uncertainties in the determination of the wavelength which
are directly translated into uncertainties in fluorescence intensities. Further requirements are revealed in
“Spectrally Resolved Macro- and Microfluorometry”
section.
Fluorescence quantum yield standards are used as
reference for the relative determination of the fluorescence quantum yield of an analyte. Accordingly—
contrary to fluorescence intensity standards that relate
chemical concentration to instrument response—they are
typically not based on the same fluorophore as the sample and is not mandatory matching of spectra. The most
stringent requirements are reliable fluorescence quantum
yields and properly defined conditions such as e.g. matrix, oxygen concentration, temperature, and excitation
wavelength as well as a stated uncertainty of the quantum yield. Furthermore, standard and sample should absorb and emit within comparable spectral regions, see
also “Spectrally Resolved Macro- and Microfluorometry”
section.
The working principle of the majority of fluorometric reference materials that relates chemical concentration
to instrument response is the comparison of the spectral
radiance/fluorescence intensity of a standard and a sample to quantify the number of fluorophores in the sample.
Though often overlooked in fluorometry, this is a common
approach used in many different analytical techniques.
Resch-Genger et al.
This type of intensity standard typically relies on the same
fluorophore(s) as to be quantified. However, due to the inherent sensitivity of the spectroscopic properties of the
majority of chromophores to their microenvironment, this
type of standard has to be handled with care. To eliminate
errors in quantification, the chromophore to be specified
and the standard have to be in the same microenvironment
to guarantee identical fluorescence spectra, molar absorption coefficients, and fluorescence quantum yields, respectively. This is for instance fulfilled for many applications
of high performance liquid chromatography (HPLC) with
fluorescence detection where the fluorescence intensities
from free, i.e. unbound fluorophores in solutions of identical or at least very similar composition are compared
[54,55]. On the other hand, for a set of fluorophores in solution and a set of immobilized fluorophores, e.g. bound
to or incorporated into a microparticle, the absorption
and fluorescence spectra, molar absorption coefficients,
and especially the fluorescence quantum yields can differ
significantly [13]. If not properly considered, these differences lead to considerable errors in fluorophore quantification. As an approach to overcome these problems, the
quantification of the fluorescence intensity of samples is
often performed in units of molecules of equivalent soluble fluorochrome (MESF) [56,57]. This type of reference
system, that circumvents differences in the molar absorption coefficients and fluorescence quantum yields of the
standard and the sample, however, relies on matching fluorescence spectra and the sample and is not designed to
derive the absolute number of fluorophores in the sample.
Accordingly, standard-related contributions to the overall measurement uncertainty are linked to the degree of
spectra matching of the standard and the sample [32].
This is also true for standards for the characterization
of material-related autofluorescence which are requested
by manufacturers of materials relevant for fluorescence
analysis like e.g. optical glasses, crystals, and ceramics
employed for the fabrication of optical components and
supports/sample containers.
For standards employed for the characterization of
the day-to-day performance and long-term stability of fluorescence instruments, a close match between the standard
and routinely measured samples is not mandatory. However, these standards should be measured/measurable with
routinely used instrument settings to guarantee the reliability of the instrument performance under applicationrelevant conditions. Further prerequisites are sufficient
and reported stability or, for single-use standards, excellent reproducibility. Contrary to statements in the literature, per definition, there are no standards but only
methods, procedures or guidelines—preferably in combination with suited materials—to measure the range of
Quality Assurance in Fluorometry and Suited Fluorescence Standards
linearity of fluorescence instruments, their dynamic range
and the limit of detection [29,30]. For instance, the determination of the former requires a method for the defined physical or chemical variation of the amount of
light reaching the detector [29,40]. Additionally, recommendations on accordingly tested materials are desired
that enable such measurements without material-specific
contributions [40].
In addition to the standards discussed, for other fluorescence instruments like e.g. fluorescence microscopes,
microarray scanners, and imaging systems, further reference materials are needed for instrument characterization and IPV. As is discussed in the sections dedicated to
these techniques, requested are for instance standards for
the determination of the spatial resolution, the correction
for the heterogeneity of illumination and for detection
within a single field, i.e. shading or flat field correction,
and the characterization of the size of the illuminated
volume.
APPLICATION-SPECIFIC NEEDS FOR
IMPROVED COMPARABILITY AND
RELIABILITY OF FLUORESCENCE DATA
Frequently used fluorescence techniques include
macro- and microfluorometry often combined with measurements of fluorescence spectra, fluorescence lifetimes,
and fluorescence (de)polarization, fluorescence microscopic techniques, and flow cytometry as well as fluorescence imaging in vivo. Fluorescence detection is also commonly employed for analytical separation techniques like
for instance HPLC, capillary electrophoresis (CE), and
(high performance) thin layer chromatography (HPTLC;
TLC), and for bioanalytical methods like rt-PCR used for
DNA analysis and DNA quantification as well as for the
microarray technology. However—despite of the broad
and ever increasing use of fluorescence techniques—to
the best of our knowledge, there exist no overall accepted
procedures for instrument characterization and IPV at
present. Different manufacturers of steady-state fluorometers recommend the so-called Raman test [58,59] for
the control of the instrument’s long-term stability and
sensitivity and the determination of the wavelength accuracy via scanning of the emission lines of the instrument’s Xenon excitation source or the transmission minima in the spectrum of a solution of holmium perchlorate.
These tests are often integrated into the software of instruments, and the recommended materials and required
accessories are available from the instrument manufacturers, in some cases even with a SOP for use. More
complex procedures recommended and commonly per-
343
formed by service engineers include the determination of
the wavelength accuracy with pen-type atomic discharge
lamps and the characterization of the spectral responsivity
of the emission channel with a calibrated lamp, i.e. typically an instrument-adapted integrating sphere-type spectral radiance transfer standard. The availability of such
procedures is limited to few types of instruments, mostly
steady-state fluorometers. For other techniques like e.g.
fluorescence microscopy no such recommendations are
available. Accordingly, in the following sections, the need
for procedures/guidelines and standards for the characterization of fluorescence instruments and performance of
fluorescence measurements is illustrated for selected fluorescence techniques.
Spectrally Resolved Macro- and Microfluorometry
Uncertainty of Fluorescence Measurements
The uncertainty [60] of fluorescence measurements
relies on parameters such as e.g. the stability of the excitation light source and means for correction of its fluctuations, the stability and linearity of the detection system(s)
and the detector’s sensitivity to polarization effects, the
occurrence of photobleaching, background fluorescence,
scattering and reflections, the measurement geometry,
sample format and reproducibility of sample positioning
as well as the type of sample and sample-handling steps,
respectively. Influences of the latter two factors are not
further discussed. As a rule of thumb, for cuvette-type fluorescence measurements of transparent dilute solutions,
the fluorescence-based uncertainty is ca. 2%. For measurements of solid films with a routine fluorometer, the
uncertainty is increased, mostly due to enhanced uncertainties related to sample positioning. For fluorescence
measurements with 96-well microtiter plates, a well-towell reproducibility of ca. 2% can be achieved under
optimum conditions, i.e. with well-aligned instruments,
well-chosen microtiter plates, and a flat meniscus for top
reading. The effect of the meniscus, that depends on the
surface tension of the solvent, solvent–well interactions,
and the volume of the analyte solution, is illustrated in
Fig. 4 for top reading. Figure 4 compares the uncorrected
mean emission spectra and the respective standard deviations (SD) obtained for two solutions of organic dyes,
i.e. coumarin 153 and quinine sulfate which emit within
the spectral region of 400–600 nm and were dissolved in
ethanol and 0.1 M HClO4 , respectively. For each measurement, the same black microtiter plate from NUNC
and a volume of 270 µl of the pipetted solution per well
were used and the standard deviation was derived from 32
wells. The smaller SD of 3% resulting for the ethanolic
344
Fig. 4. Uncorrected mean emission spectra of two organic dyes and the
respective standard deviations (SD) measured with a common microplate
reader using a black 96-well microplate from NUNC and a pipetted
volume of 270 µl, respectively. For each dye, spectra are averaged over
32 wells. Left: coumarin 153 in ethanol, right: quinine sulfate dihydrate
in 0.1 M perchloric acid.
Resch-Genger et al.
Fig. 5. Comparison of the uncorrected (dashed lines) and spectrally
corrected (symbols) emission spectra of a fluorophore-labeled polymer
film measured at two different settings of the emission polarizers, i.e. 0◦
and 90◦ . The bump in the uncorrected spectrum obtained for 90◦ results
from diffraction effects of the monochromator gratings used.
Fluorescence Standards Suited for Spectral Correction
coumarin 153 solution (left), which shows a rather flat
meniscus, compared to the SD of 8% obtained for quinine
sulfate dihydrate (right) exemplary demonstrates the influence of solvent-related meniscus effects.
Need for Spectral Correction of Fluorescence Spectra
The influence of the instrument-specific spectral
characteristics and the according need for spectral correction is exemplary highlighted for steady-state fluorometry in Fig. 5 comparing the uncorrected and corrected
emission spectra of a typical fluorescent label used for the
derivatization of amino groups, here at the surface of a
polymer support. The measurements were performed at
two different settings of the emission polarizers, i.e. 0◦
and 90◦ . As is evident from Fig. 5, the uncorrected emission spectra differ considerably in fluorescence intensity
and shape for both polarizer settings. The bump in the
uncorrected curve obtained for an emission polarizer set
to 90◦ arises from diffraction effects or so-called Wood
anomalies of the monochromator gratings of the instrument. Moreover, both uncorrected spectra clearly deviate
from the corresponding corrected spectra that are almost
identical. These findings demonstrate the need for spectral
correction to obtain comparable instrument-independent
spectra and the potential of spectral correction to eliminate instrument-related polarization effects that can distort
uncorrected spectra.
Spectral correction of fluorescence data provides the
basis for the comparability of fluorescence data across instruments, laboratories, and over time as well as for the
traceability of fluorescence measurements as illustrated
in Figs. 1 and 5. With the increasing application of fluorescence instruments with real spectral resolution like for
instance detectors for chromatography, that record fluorescence spectra, as well as modern types of fluorescence
microscopes and the according desire to use fluorescence
spectra for species identification, spectral correction, and
the availability of reliably corrected fluorescence spectra
are gaining importance for many other fields of application. Furthermore, spectral correction is mandatory for
fluorescence methods that rely on the comparison of two
fluorophores with different absorption and emission features or measured at two different wavelengths.
The traceable correction of fluorescence data for
instrument-specific wavelength dependences can be realized via measurement of emission and excitation correction curves in relative intensities [40] either with physical
transfer standards, i.e. calibrated standard lamps and detectors [42], or certified emission and excitation standards.
Due to their ease of use and resemblance to typically
measured samples, the latter approach is recommended
for the broad community of users of fluorescence techniques. In addition to the previously discussed general
requirements on fluorescence standards, and the need for
broad and unstructured fluorescence spectra, this type of
Quality Assurance in Fluorometry and Suited Fluorescence Standards
345
spectral fluorescence standard should have moderate to
high fluorescence quantum yields independent of excitation or emission wavelength and dye concentration to
enhance the signal-to-noise ratio and to reduce the influence of stray light, solvent emission, and fluorescent
impurities on the shape of the standard’s fluorescence
spectrum [40,47]. Minimum overlap between absorption
and emission is important to minimize inner filter effects. A nearly isotropic emission is the prerequisite to
guarantee negligible additional uncertainties under measurement conditions that can dispense with or for instruments that lack polarizers. Furthermore, to cover the
UV/Vis/NIR spectral region, sets of spectrally tailored
standards are requested in combination with a tested procedure for the determination of the overall spectral correction curve from the measured and provided corrected
spectra [40,46,47].
Measurement of Fluorescence Intensities
To circumvent the use of standards, absolute measurements of fluorescence intensities or quantum yields
can be performed. However, in addition to the consideration of the instrument’s spectral characteristics, this
requires the determination of the absolute fraction of incident photons per time unit in the sample that are absorbed by the analyte and the collection efficiency of the
instrument, which together control the fraction of the fluorescence photons detected. This renders absolute measurements very challenging. Also, each instrument has to
be accordingly characterized to establish the comparability of fluorescence intensities. On our opinion, this is not
practicable on a broad level, i.e. for each instrument used
for quantitative fluorometry.
For the broad community of users of fluorometry, a
purpose-fit approach to quantitative fluorometry including the determination of fluorescence quantum yields is
the application of fluorescence intensity standards. For
these applications, the influence of spectra matching and
spectral correction is exemplary illustrated in Fig. 6 for a
series of commonly used fluorescent dyes, the corrected
emission spectra of which are shown in the lower part
of Fig. 6. The upper part of Fig. 6 depicts the integral
fluorescence intensities of these dyes that have been determined relative to quinine sulfate dihydrate [61] taking
the areas from the respective integrated uncorrected and
corrected emission spectra, respectively. All the measurements were performed at the same excitation wavelength
and bandpass. The emission spectra were backgroundcorrected and have been converted from radiometric to
photonic quantities prior to integration [40]. As is evident
Fig. 6. Comparison of the integral fluorescence of several organic dyes
obtained from uncorrected and corrected emission spectra, respectively,
using quinine sulfate dihydrate in 1 N sulfuric acid as fluorescent standard. Excitation was at 370 nm. Bottom: Corrected emission spectra
of α-NPO (short dash-dotted), coumarin 47 (solid line), coumarin 102
(dotted), coumarin 153 (dash-dotted), rhodamine 101 (dashed line), and
quinine sulfate dihydrate (-◦-).
from Fig. 6, the magnitude of these now and then considerable deviations between uncorrected and corrected
data depend on the shape of the wavelength dependence
of the spectral responsivity of the instrument used, see
also Fig. 1, and for a given standard accordingly on the
spectral region of the dye’s emission and the width of
its emission spectrum. For each sample-standard pair,
uncorrected fluorescence spectra can be compared with
a reasonable uncertainty only for species with identical
or at least closely matching fluorescence spectra. This is
the reason why for fluorescence techniques using integral measurements of relative fluorescence intensities for
quantification purposes like e.g. flow cytometry, spectra
matching is recommended [32,56]. Due to the instrument
specifity of spectra matching of a sample-standard pair,
its dependence on dye microenvironment, and the ever
increasing variety of fluorescent labels and fluorophorecontaining systems, general solutions to quantitative fluorescence analysis with small uncertainty are not very
346
realistic. Fluorescence-based uncertainties of ca. 2% can
be realized only for simple cases where the fluorescent
analyte and the standard are identical and located in the
same or in a very similar microenvironment. This is often met e.g. for HPLC analysis of fluorescent analytes
such as polycyclic aromatic hydrocarbons (PAH) [54,55].
However, for more complex systems, a considerable increase in measurement uncertainty of up to a few tens
of percent may occur. This is further detoriated especially by the dependence of the fluorescence quantum
yield on microenvironment. To meet this challenge, it
may be of interest to determine fluorescence-based uncertainties for representative examples, i.e. very popular
labels and matrices, with varying degree of spectra matching and well-characterized microenvironment-induced
changes in molar absorption coefficients and fluorescence
quantum yields.
Fluorescence Quantum Yield Standards
From the material side, the sensitivity of fluorescence
analysis depends on the molar absorption coefficient of the
sample/analyte(s) at the excitation wavelength and the fluorescence quantum yield. Accordingly, the availability of
fluorescence quantum yields of (bio)analytically relevant
dyes in application-relevant matrices under applicationrelevant conditions and their accurate determination are
of considerable importance. The uncertainty of the measurement of relative fluorescence quantum yields is governed by several factors. Major contributions arise from
the uncertainty or lack of spectral correction, see Fig. 6,
temperature-related effects, uncertainties inherent to absorption measurements and dilution of solutions as well
as from the uncertainty of the fluorescence quantum yield
of the chosen standard, respectively. With respect to the
latter, it is critical that literature data on fluorescence quantum yields are often not very reliable with deviations of up
to 50% being not uncommon and uncertainties typically
missing. Accordingly, there is still a considerable need for
fluorescence quantum yield standards for the UV/Vis/NIR
spectral region, the fluorescence quantum yields of which
have been determined with a stated uncertainty with a
traceably characterized setup. To eventually overcome
these problems, currently at BAM, a reference fluorometer
is been built for the dissemination of absolute fluorescence
quantum yields. Furthermore, not only well-characterized
standards with high fluorescence quantum yields are
desired but also systems of low fluorescence quantum
yield. This takes into account that not only the absorption and emission spectra of standard and sample should
lay within a similar spectral region but also the size of
their quantum yields should be preferably close to avoid
Resch-Genger et al.
problems related to nonlinearities of the detection system
or dilution errors.
Standards for the Characterization
of Material-Related Autofluorescence
The inherent sensitivity of fluorescence imposes
strong requirements on materials suited for the fabrication of optical components for fluorescence instruments
as well as supports for fluorescence analysis. This eventually affects quality control by materials suppliers as is
exemplary revealed for glass-type materials. Generally,
materials made for applications in optical technologies
should show minimum autofluorescence on excitation at
wavelengths typically used, i.e. LED, and laser (diode)
wavelengths as well as at 365, 248, and 193 nm. Furthermore, they need to be stable under application-relevant
conditions, e.g. at photon fluxes in the range between
µW/cm2 for spectroscopy/microscopy and kW/cm2 for
microlithography, and should not form fluorescent photoproducts [62,63]. With respect to proper quality control
for glass-type materials, it has to be taken into account
that their optical properties and (photo)stability are influenced by (fluorescent) impurities/dopants, the presence
of point defects such as color centers that originate for
instance from impurities such as RE elements, as well as
the matrix. Also, the glass manufacturing process itself
plays a role, i.e. redox equilibria and eventually incorporation of dopants into the glass matrix [64–66]. Accordingly, material-related autofluorescence needs to be
specified under application-relevant conditions, e.g. excitation wavelengths and photon fluxes [67], and on a
batch-to-batch basis, see also Fig. 3. Techniques that are
increasingly used for this purpose are conventional steadystate, laser-induced, and time-resolved fluorometry [68–
74]. For instance, a standard procedure for quality control
used by suppliers of optical glass is the measurement of the
glass emission at an excitation wavelength of 365 nm and
comparison of the integral emission in the wavelength region of 400–700 nm to that of reference samples like SF1
or SF6 available from Schott AG. SF1 and SF6 contain a
defined amount of emissive lead ion chromophores, e.g.
ranging between 20 and 70%. A similar approach based
on lead-type glasses certified by the Agency of Industrial
Science and Technology (AIST) is performed within the
Japanese Optical Glass Industrial Standard (JOGIS) procedure applied by Japanese manufacturers [75]. Glasses
of low emission show an integral photoluminescence well
below 5%.
Despite of the considerable importance for optical
industry, instrument manufacturers and for the reliability of fluorescence analysis—aside from these lead-based
Quality Assurance in Fluorometry and Suited Fluorescence Standards
standards—up to now, there are no suited solid reference
materials available that enable a standardized determination of the application-relevant fluorescence properties
of glass-type materials. Desired are fluorescence quantum yield standards and/or reference materials that relate
the concentration of typically encountered fluorescent impurities or dopants to fluorescence intensity, standards
to quantify the acceptable level of autofluorescence for
quality control as well as standards for the fluorometric
quantities luminescence lifetime and (de)polarization. In
addition to the requirements on quantum yield and fluorescence intensity standards, of special importance for
this application are the shape and size of potential reference materials and the respective measurement geometry,
e.g. 0◦ /90◦ and front face. Also the luminescence quantum yields of the analytes of interest have to be taken into
account. Luminescence quantum yields are in the range
of 60% for RE-doped optical glass and CaF2 due to the
inner-shell nature of the optical transitions involved, reach
up to 95% for color centers [76], and are below 10% for
optical transitions from the ns2 levels (n = 1, 2, 3, etc.)
as is found for instance in SbO3 , AsO3 or ZnO. Suited
luminescence lifetime standards should cover the ns- to
ms-time domain. In a first step, to help manufacturers of
optical materials and fluorescence instruments to improve
fluorescence-based quality control, a set of reference materials covering the most important application-relevant
problems related to autofluorescence from glass-type materials needs to be defined, certified, and made available.
Furthermore, internationally accepted guidelines for the
determination and specification of autofluorescence are
increasingly requested. This will eventually contribute to
an improved performance of fluorescence instruments and
reliability of fluorescence analysis.
347
etry is used to detect fluorescently labeled intracellular
constituents like DNA or messenger RNA.
One essential feature of flow cytometry is the analysis of single particles at a high throughput of up to
100 kHz. For this purpose, the sample fluid containing
the suspended particles is injected into a flow cell together with a second fluidic stream, the sheath or mantle
stream. This results in hydrodynamic focusing of the sample stream to diameters of typically 5 µm that correspond
approximately to the size of human blood cells. Hence,
the cells are forced to cross one or several laser beams,
oriented perpendicularly with respect to the direction of
the flow, in single file. When passing the laser beams,
for each cell, light scatter in forward and sideward direction and laser-induced fluorescence is simultaneously
recorded. Generally, the pulse heights of the signals are
measured. Fluorescence is observed in up to four different
optical bandpasses or channels corresponding to different
antibodies and different fluorophores, respectively.
In most cases, relative fluorescence intensities are
measured in flow cytometry to divide cell population into
two species, one of which reacts with a selected monoclonal antibody and hence is termed positive. The other
population that does not react with the selected antibody
is referred to as negative. In Fig. 7, a typical histogram
obtained with the antibody CD4 (Cluster of Differentiation 4) labeled with the fluorophore PE (phycoerythrin)
is presented which features two populations, i.e. CD4positive (CD4+ ) lymphocytes and CD4-negative lymphocytes [77]. The difference or ratio of the distributions
of fluorescence intensities between these two cell subsets strongly depends on several factors. This includes the
instrumental characteristics, the spectroscopic properties
Flow Cytometry
Flow cytometry is a technique frequently used in
clinical medicine to support diagnoses or prognoses of
different diseases such as leukaemia, HIV infection or
immune paralysis, i.e. sepsis. To derive the information
required for a medical decision, cells of a blood or bone
marrow sample as well as cells isolated from a biopsy
are distinguished according to their physical and biological properties. Subsequently, concentrations or relative
concentrations of subpopulations of cells are determined.
Apart from cell differentiation based on light scattering in
forward and sideward directions, staining of the surface of
target cells by fluorophore-labeled monoclonal antibodies has evolved as a powerful tool for the identification
of subpopulations of cells, e.g. T-helper and T-suppressor
lymphocytes in AIDS diagnosis. In addition, flow cytom-
Fig. 7. Histogram of the frequency distribution of stained white blood
cells. Phycoerythrin (PE) labeled antibodies CD4 were used to identify
CD4-positive T-lymphocytes. Included in the figure is the discrimination
threshold to distinguish positive and negative cells. The histogram was
provided by V. Ost, PARTEC GmbH.
348
of the fluorescent molecules attached to the antibody,
and cell-specific biological features like autofluorescence
from intracellular constituents and non-specific binding.
To account for contributions arising for example from
autofluorescence and from non-specific binding, control
experiments (variation of the spectral irradiance of exciting laser beams, isotype control to determine non-specific
binding, titration of antibody concentration) are required
to optimize the difference between positive and negative
populations, i.e. the signal-to-background ratio. To facilitate the comparison of the results from different instruments, to compare different blood samples, and to accomplish a day-to-day control of the flow cytometer, stained
microbeads are frequently used. A typical histogram displaying the relative frequency of particles of different fluorescence intensity is depicted in Fig. 8. The number at
each peak corresponds to MESF units. The MESF number
does not represent the true amount of dye molecules, but
indicates the brightness, i.e. the product of the molar absorption coefficient and the fluorescence quantum yield, of
the fluorescent species in solution [13,39]. MESF particles
are available with different fluorophores like for instance
fluorescein (FITC), PE or allophycocyanin (APC) and are
routinely used to align the optics of flow cytometers, to
set the detection threshold as well as to check the sensitivity and the linearity of the instruments. Alternatively,
the linearity and resolution, i.e. the width of the intensity
distribution of flow cytometers can be monitored with
biological material, e.g. DNA-stained trout or chicken
erythrocytes.
Fig. 8. Histogram of differently stained calibrations beads (SPHEROTM
Rainbow Fluorescent Particles). The fluorescence intensities given in
arbitrary units can be calibrated in MESF units. The corresponding
MESF numbers labeling each peak indicate its respective center. The
fluorescence was observed in the FITC detection path of a Cytomation
MoFlocell sorter.
Resch-Genger et al.
To improve quality assurance in flow cytometry, the
spectroscopic features of existing calibration (MESF) particles should be better characterized. This implies for example the determination of the absorption spectrum and
the molar absorption coefficients at the wavelengths commonly available in flow cytometers as well as the corrected emission spectrum and fluorescence quantum yield
in phosphate buffered isotonic solution. The availability of
such data would allow users of flow cytometry to obtain a
spectral correction of the instrument and to determine the
crosstalk between different fluorescence detection paths,
the correction of which is termed fluorescence compensation. This eventually considers future developments in
flow cytometry that might result in an improved spectral
resolution from presently 20–30 to 1–2 nm. Accordingly
certified MESF beads are suited to calibrate such instruments which allow to record emission spectra.
Apart from relative measurements, quantification of
fluorescence intensity in flow cytometry and subsequent
deduction of the significant biological quantities, i.e. the
number of specific antibodies on the cell surface (antibody binding capacity, ABC) or the number of proteins
within the cell, would possibly allow to improve medical
diagnosis. For example, standardized immune monitoring, which includes the determination of ABC for different antibodies, has been recently applied [78] to derive
criteria for the prediction of infections in risk patients after cardiopulmonary bypass surgery. For monocytes, the
expression of HLA-DR (human leukocyte antigen) epitopes correlates with their activation and can be applied
to monitor patients with septic shock [79]. To determine
the number of antibodies, the QuantiBRITEprinciple is
used [80]. For this calibration procedure, PE is chosen
to label antibodies and calibration beads as the influence
of the microenvironment on its spectroscopic properties
is not as pronounced as for other fluorophores. Another
advantage of PE is its high molecular weight and its correspondingly large size, which allows to control that only
one fluorochrome is bound to each antibody. Apart from
quantification of PE fluorescence, the expression of enhanced green fluorescent protein (EGFP) of transfected
T-cells has been determined by calibration of the fluorescence of cells to an EGFP standard of validated MESF
units [81].
Due to the inherent sensitivity of fluorescence to dye
microenvironment, for each chromophore, diverse environmental conditions can cause significant variations in
the brightness and spectral position of the absorption and
emission bands of the stained cells. The classical example
is the pH dependence of the fluorescence quantum yield
of frequently used fluorescein which limits its use. Hence,
to correct for such variations, any standard used to derive
Quality Assurance in Fluorometry and Suited Fluorescence Standards
biological quantities has to mimic the fluorescently labeled cells as closely as possible. However, because of the
complexity of bacteria or cells, it is challenging to achieve
matching of absorption and emission spectra of standard
and biological sample. In addition, it is ambiguous to control labeling-induced changes of the spectral features, the
molar absorption coefficient, and especially the fluorescence quantum yield. To assure the traceability of such a
standard and to quantify its spectral properties, comparison with corresponding stained cells is mandatory, for
example by employing cell sorting and subsequent single molecule bleaching under microscopic control. This
technique would allow to determine the properties of fluorophores in situ and to account for the influence of the microenvironment. In addition, to derive the biological influencing quantities, colocalization should be applied to the
cells under investigation to distinguish specific and nonspecific binding of the fluorophore-labeled monoclonal
antibodies.
Fluorescence Microscopy
As a tool in microscopy—in addition to absorption methods—fluorescence provides a number of already mentioned possibilities such as e.g. selectivity of
communication via different parameters like excitation
and emission wavelength, fluorescence lifetime, and fluorescence (de)polarization [5,82–85]. Furthermore, with
confocal laser scanning microscopy and wide-field microscopy with spatially modulated excitation, the theoretical limits of spatial resolution that are determined by
the numerical aperture of the objective and the excitation wavelength can be realized in practice. In addition,
labeling or probing of biological structures with fluorescent reporters allow their indirect visualization even
when the size is far below the optical resolution limit.
Accordingly, confocal and wide-field fluorescence microscopy have been developing into some of the most
powerful and commonly used methods in medicine, forensics, and bioanalysis. Founded on these techniques, further specialized imaging and non-imaging methods like
two-photon fluorescence microscopy (2P), F¨orster or fluorescence resonance energy transfer (FRET), total internal
reflection microscopy (TIRF), fluorescence lifetime imaging microscopy (FLIM), fluorescence recovery after photobleaching (FRAP), fluorescence loss in photobleaching (FLIP), fluorescence intensity (FLINT), fluorescence
anisotropy (FA), time-resolved energy transfer (TRET) as
well as fluorescence correlation spectroscopy (FCS) have
been become very popular. Typical applications of fluorescence microscopy techniques are investigations of fixed
(dead) samples like immunofluorescence studies and in
349
situ hybridization measurements of DNA sequences, live
cell imaging with measurements of the structure, dynamics, and potential (maps) of membranes and organelles,
measurements of cell structures, organization, and function, e.g. determination of intracellular pH and physiologically important ions like e.g. Ca(II), Na(I), K(I), and
Mg(II) [86] as well as studies of protein structure and
dynamics [5,82–84,87]. A broad variety of these microscopic techniques do not require quantification. However,
methods for the study of biological functions are gaining importance that demand the precise quantification of
the concentration of analytes by measuring fluorescence
intensity in the 3D space over time (4D). Furthermore,
as some experimental approaches or methods like FRET
require quantitative spectrally resolved measurements of
two fluorophores simultaneously, this task can result in
5D data sets. The need for quantification, however, is
poorly met at present as commercial imaging instruments
are designed mostly for high image quality, i.e. low background signals and image distortions, high light throughput, and good detection efficiency, but not primarily for
quantitative measurement of fluorescence. Accordingly,
this situation is reflected by the commercially available
standards.
Due to the complexity of the instrumentation used for
confocal and high-end wide-field imaging microscopy,
there is an urgent need for easy-to-use standards and
simple procedures for the characterization of these instruments, control of instrument specification, day-today and long-term performance as well as to enable the
comparability—and in some cases the combination of microscopic data—across instruments and laboratories. This
can be seen as the first necessary step towards calibration
procedures and standards fulfilling the up-coming need
for signal quantification in many fields of microscopic
imaging. The demand for standardization of instrument
characterization and measurement procedures is further
enhanced due to the increasing use of fluorescence microscopy in areas like medical diagnostics. Accordingly,
in this section, the general need for and requirements
on standards suited for fluorescence microscopy are discussed without consideration of the specific demands of
the broad variety of fluorescence microscopic techniques
available.
In conventional and confocal fluorescence microscopy, the recorded signal is affected by the samplespecific amount of fluorescent molecules with a certain
molar absorption coefficient and fluorescence quantum
yield present in a given microscopic volume, the instrument, and autofluorescence. The latter can arise
for instance from the sample, the instrument’s optics
and/or from an immersion medium. (Time-dependent)
350
instrument-specific effects are linked to the wavelengthdependent spectral irradiance at sample position, the
wavelength- and polarization-dependent light collection
properties and aberration of the microscope, and the
wavelength- and polarization-dependent spectral responsivity of the detection system [88]. To obtain instrumentindependent and hence comparable data, these instrument
characteristics need to be measured. For the quantification of the recorded signals, the linearity of the detection
system has to be determined and either standards closely
matching the sample/analyte to be quantified or appropriate characterization procedures are required [89]. According to the broad community of fluorescence microscopists
[90], instrument-type standards—preferably in combination with guidelines for instrument characterization—are
desired to characterize the size of the illuminated volume,
i.e. the point-spread function [91], the instrument’s spatial
(x, y) resolution, the spectral irradiance/excitation intensity reaching the sample, the homogeneity of the sample
illumination, the field flatness, z-distance and z-resolution,
the spectral resolution as well as the day-to-day and
long-term instrument performance, respectively. Moreover, application-specific fluorescence intensity standards
or so-called reference standards are requested [92] and
internationally agreed methods/protocols in combination
with suited systems to determine the range of linearity of
the detection system [93]. Additionally, guidelines including materials to measure limits of detection are of interest,
for instance to compare the sensitivities of different instruments. Furthermore, with the increasing spectral resolution offered now by the majority of modern commercial
confocal microscopes, spectral fluorescence standards are
gaining importance [94].
Standards for the characterization and calibration of
fluorescence microscopes have to meet additional requirements compared to standards for macroscopic fluorescence spectroscopy. This is due to the strongly increased
spectral irradiances at the sample position that exceed
those used in conventional (macro)fluorometry often by 4
or more orders of magnitude [17]. Accordingly, frequently
enhanced photodecomposition is observed, especially in
the case of laser illumination typically used for confocal
scanning techniques. The strongly improved spatial resolution imposes more strict requirements on the homogeneity of the chromophore distribution within the standard.
Size and shape of a standard are by far more important
as microscopic measurements usually require changes in
field aperture and magnification. Only dimension-adapted
microscopic standards with a well-defined shape permit
a correct intensity/volume relationship. Furthermore, the
spectral radiances and spectral characteristics of the standard and the sample need to be sufficiently similar.
Resch-Genger et al.
At present, there is a broad variety of fluorescence
standards, mostly of instrument-type, commercially available. Instrument standards can be of chemical, i.e. organic
or inorganic nature, or of physical type. Standards, which
relate instrument response to analyte concentration, commonly rely on organic fluorophores either incorporated
within or bound to the surface of a stable, non-fluorescing
host of well-defined shape. Other approaches are microdroplets of fluorescent solutions [92] or microcapillaries
filled with fluorophore solutions [17,95,96]. Typical instrument standards are for instance inorganic ion-doped
glass materials such as uranyl-containing glasses [97,98]
that are of good photo-stability but no longer commercially available, or inorganic ion-doped fibers [99] developed for the calibration of the day-to-day instrument
performance. Inorganic ion-doped glasses have only restricted application as reference standards for certain
analytes with sufficiently similar excitation and emission spectra. Fluorescent microspheres [100–102] are of
widespread use not only as instrument-type standards to
check on instrument alignment, sensitivity, and stability as
well as spectral separation (multicolored particles) of both
conventional microscopes and laser scanning systems but
also as internal as well as external fluorescence intensity
standards. Some examples for measurements with fluorescent microspheres are shown in Fig. 9. The dually labeled
beads (FocalCheck, Molecular Probes) shown are applied
for alignment purposes (parfocality, xy pixel-shift with
color) in confocal and wide-field fluorescence microscopy
when two or more excitation/fluorescence channels have
to be acquired from one location in the 3D space. They
allow to adjust for pixel-shift free recordings in the 3D
space, see Fig. 9C and D, respectively. However, these
microspheres can act as an additional optical lens element
in the optical path, leading to severe optical distortions in
the 3D space as follows from Fig. 9B. The applicability
of such particles is further limited by the fluorophore’s
stability and often by the mounting medium.
Other suggested standards include concentrated solutions of fluorophores like fluorescein and rhodamine
dyes on a regular slide [103] for shading or flat field correction, fixed fluorescent cells [104] or fluorescent polymers,
which can, however, suffer from spatial non-uniformity
leading to inhomogeneous emission [105], immobilized
particle arrays [106] or wax films doped with fluorescent dyes [107]. Recently, organic and inorganic systems
containing uniformly dispersed luminescent nanocrystals
at various concentrations have been suggested as potential fluorescence standards for microscopy [108]. However, aside from the generally advantageous spectroscopic
properties of quantum dots, i.e. size-tunable broad absorption and narrow emission bands, and comparatively high
Quality Assurance in Fluorometry and Suited Fluorescence Standards
351
Fig. 9. Focal check beads usable for parfocal alignment and shift free multi-channel
acquisition in confocal and wide-field fluorescence microscopes. The beads are labeled
with one color throughout and with a second color on the outer shell. (A) Overlay of a
three-channel image acquired sequentially with 364, 488, and 543 nm excitation (mixture
of beads F-7237 and F-7238, Molecular Probes Inc., Eugene, USA). (B) Orthogonal x–z
and y–z cross-section view from a two-channel z-stack (excitation at 364 and 488 nm). It
can be seen, that the bead acts as an optical active element, leading to a distorted intensity
profile of the green colored shell. (C) x–z cross-section view of a bead squeezed between
two microscope slides, well-aligned collimator for UV/Vis illumination. (D) x–z crosssection after misaligning the collimator, which results in a shift of one color in relation to
the other. All images were recorded with a 63×/1.2 water immersion lens on a LSM 510
Meta confocal microscope (Carl Zeiss GmbH, Jena, Germany).
fluorescence quantum yields in the long wavelength region [109], the suitability of such systems as standards
requires a very narrow and reproducible particle size distribution and a time- and illumination-independent brightness [110]. The latter implies the overcome of the often observed increase in fluorescence intensity on illumination
or so-called photobrightening [111]. Also physical-type
standards like for instance lamp-based calibration assemblies have been reported [94]. An intriguing approach is a
test slide that contains a built-in regulated light-emitting
diode in feedback configuration with a PIN photodiode
and different pinholes to mimic the emission characteristics of fluorescent cells [112]. This assembly allows the
determination of the linearity of the detection system of
a microscope. Another multifunctional standard is a microscope test slide that contains side by side a photodiode
for power and pulse length measurements and a detector
for wavelength measurements [113].
Despite of the enormous progress in fluorescence
microscopy and the broad variety of fluorescence standards suggested, up to now, the suitability of the available
materials as well as the need for improved standards for
the characterization of fluorescence microscopes and for
quantification purposes is still under debate [90]. This is
also true for the demand of internationally accepted protocols for instrument calibration, control of instrument
specifications, and IPV as well as signal quantification.
Disadvantages of some of the available or recommended
fluorescence standards that need to be considered and
eventually overcome, are high, not sample comparable
fluorescence intensities emitted even at low excitation intensities which lead to enhanced calibration uncertainties caused by nonlinearities of the detection system or
stray light effects within the instrument. Critical can be
also a non-adequate thickness of the fluorescence standard that renders the identification of the appropriate focal plane for the calibration procedure difficult. Also,
the effect of the coverslip has to be taken into account
for the design and selection of a suited standard. Measurements of a standard without a coverslip may be erroneous due to distortion of the measured signal in the
light path, as most of the objectives for biological imaging are designed, i.e. calculated, for use with a coverslip of 170 µm thickness. However, for objectives made
for application without a coverslip—water dipping lenses
employed in neurobiology or lenses for use in material
science—coverslip-free standards are needed. Furthermore, severe optical distortions due to strong refractive
352
index mismatches between standard and immersion media can be problematic and should be avoided when the
3D dimension of the standard is of importance. Other
standard-related problems and sources of error can be an
insufficient or not reported stability under laser illumination that leads to photodecomposition of the fluorophore
and accordingly aging-inherent intensity and/or spectral
variations and/or an inhomogeneous distribution of the
chromophore within the matrix for solid standards such as
color slides.
Resch-Genger et al.
To the best of our knowledge, long-term stability
problems have been encountered for almost all types of
potential standards for fluorescence microscopy. The influence of such photoinduced effects is exemplary illustrated in Fig. 10 for colored, fluorescent plastic slides
that are used in confocal and wide-field microscopy for
spectrally resolved measurements, spectral calibration as
well as adjustment or control of the homogeneity of the
illumination. These slides may be also suited for the evaluation of (fluctuations of) the excitation light intensity or
Fig. 10. Fluorescence emission spectra and photobleaching of plastic color microscope slides from different suppliers (Deltavision (DV), MM).
(A) Emission spectra of the color slides recorded on a LSM 510 Meta (Carl Zeiss GmbH, Jena, Germany) confocal microscope with the appropriate
primary dichroic beamsplitter (DC) and a 63 × /1.2 water immersion lens. The bandwidth of the spectral Meta unit was set to 10.2 nm. A distortion
of the real emission spectrum as a result of the optical properties of the DC was not taken into account. (B and C) Photo-stability/bleach resistance
of the calibration slides. The measurements of times series of images were performed as follows: Within regions of interest (ROI), marked with
solid lines, additional scans with maximum laser intensity were run to probe the bleaching characteristics of the slides. The settings for these
ROI were: 1000 scan iterations, pixel time 1.4 µs, scanning time in total for the ROI ca. 2.5 s (DAPI Blue, Ex 364 nm,) or 25 s (DV 488/519,
excitation at 488 nm). This was repeated with the DAPI Blue sample (B) every 2.5 s for 20 times and for DV 488/519 (C) every 2.5 s for 60 times.
Images of the total area shown were taken in between the high-intensity ROI excitation: Five images (DAPI Blue) and 10 images (DV 488/519).
Measurements of the emission intensity over time within the upper ROI positions result in the non-bleached curves (dashed lines), the ROIs
marked with solid lines yield the bleached curves.
Quality Assurance in Fluorometry and Suited Fluorescence Standards
the determination of the sensitivity of fluorescence microscopes. The emission spectra of these slides and the photostability studies are shown in Fig. 10A, 10B, and 10C,
respectively. The low resistance to photodecomposition in
combination with the presence of defects in the fluorescent
layer makes these slides not very attractive candidates for
other than spectral calibration tasks provided that no lightinduced spectral changes occur. To overcome problems related to photodecomposition requires either materials of
an improved or at least better characterized photo-stability
or single-use materials of excellent reproducibility.
A first step to the design of better-suited fluorescence standards for fluorescence microscopy seems to be
the development of fluorophore-doped glass- or polymerbased slides and/or slides with fluorescent coatings for
the spectral region of ca. 450 to 900 nm that show ideally little photodecomposition under application-relevant
conditions and a homogeneous distribution of a single
or a mixture of different fluorophores. The albeit stated
photo-stability should be given in terms of applicationrelevant light fluxes at typically used excitation wavelengths. A more single-use type approach could include
for instance the application of slide-type cells or capillaries with a defined optical pathlength filled with (renewable) fluorophore solutions. With respect to the desired standardization of microscopic measurements, the
application-relevant spectroscopic properties of these potential reference materials, that need to be tested with a
broad variety of commercially available fluorescence microscopes, should be preferably certified [37].
Fluorescence- and Biology-Related Problems
in the Microarray Technology
Biochips have become an increasingly important research tool for the biotechnology industry, molecular diagnostics, and related fields of gene expression analysis,
drug screening, nucleic acid sequencing, and mutation
analysis [114–119]. Such biochips consist of a plurality
of binding agents or so-called probes. Such probes can be
peptides, oligonucleotides, or up to ten thousands of single stranded DNA fragments representing a single gene or
genome that are deposited onto the surface of a solid support like a coated glass slide in the form of an array with
spot sizes in the 50 to 200 µm range. For typical DNA microarray experiments with fluorescence detection, RNA or
DNA from biological samples and from controls is isolated and—during a reverse transcription reaction—that
transcribes RNA into cDNA—labeled with two spectrally
distinguishable dyes like the cyanine dyes Cy3 (absorption at ca. 550 nm, emission at ca. 565 nm) and Cy5
(absorption at ca. 650 nm, emission at ca. 675 nm), re-
353
spectively. The labeled biomolecules are then reacted, i.e.
hybridized with the microarray thereby binding to their
complementary sequences present in the spotted probes.
After processing, the microarrays are read out with an
array reader, typically a confocal laser scanner equipped
with a He–Ne laser (543,5, 594, 612, and 632.8 nm) or
a Kr–Ar laser (488, 514, 568, and 647 nm) or less common, a solid-state laser or a white light source and a PMT
or a CCD (charge coupled device) detection system with
a lateral resolution of 5–10 µm, respectively. This step
involves excitation at two different wavelengths and subsequent recording of integral fluorescence intensities in
two detection channels, the so-called green and red channel [120]. Usually, the scanner software is able to combine signal intensities of both channels and to deliver a
false-color image. The amount of analyte bound to a spot
follows from the measured integral fluorescence intensity with abundant targets seen either in red or green and
equal amounts in yellow. This allows for the estimation
of relative expression levels of the genes from the fluorescence intensities of the spots. For the generation of reliable
gene expression data, several repetitions and controls are
necessary with the obtained expression values being statistically analyzed and normalized. Further bioinformatic
analysis, e.g. hierarchical agglomerative clustering of the
gene expression data, provides the basis for the interpretation of the alteration in gene expression of many genes
in parallel [121–123].
Microarray experiments are performed either by ratio or direct measurements comparing two RNA samples,
e.g. healthy and disease stage, or by absolute or indirect
measurements comparing one RNA sample to a reference,
e.g. standardized RNA or DNA. Accordingly, this yields
either relative expression values (ratios) or absolute expression values. To improve the reliability of the data,
suitable controls are introduced that are incorporated into
every experiment and are accordingly present on the array
and spiked into the hybridization solution. This includes
range, ratio, negative, and contamination controls as well
as a positive control and a guide-dot for orientation after scanning. As an example for both approaches, image
files of the two different experimental designs measuring direct (A) and indirect (B) gene expression values
of the plant Arabidopsis thaliana during its growth are
shown in Fig. 11, respectively. For the former, RNA from
two different stages are labeled either with Cy3 or Cy5
and hybridized to an Arabidopsis microarray. For indirect
measurement of gene expression values, the target sample is labeled either with Cy5 or Cy3, and a vice versa
labeled reference oligonucleotide that is complementary
to the PCR primers used for probe generation is spiked
into the hybridization solution and co-hybridized.
354
Fig. 11. Image files of two different experimental designs that enable
the comparison of gene expression between samples from Arabidopsis
thaliana growth development. For direct (ratio) measurement of differences in gene expression, see (A), RNA from two different developmental stages were labeled either with Cy3 or Cy5 and hybridized to
an Arabidopsis microarray, respectively. For indirect (absolute) measurement, see (B), RNA from one developmental stage was labeled
with Cy5 and the sample was co-hybridized with a Cy3-labeled reference oligonucleotide and measured as described before. The microarrays
were scanned at excitation wavelengths of 532 and 635 nm, respectively,
and read out with the respective green and red channel with a resolution
of 10 µm using a Fuji FLA 8000 microarray scanner. Only sections of
the corresponding images are shown.
The uncertainty and reliability of the results obtained
from microarray experiments depend on the biological
and the experimental variation as well as—to a smaller
extent—on fluorescence detection and data analysis. Furthermore, due to the lack of generally accepted standardized sampling protocols and biological standards to be
used in every microarray experiment, problems related to
biology and data analysis are amplified [124]. In general,
the contribution to the overall uncertainty from the biological side has the highest impact due to genetic and
environmental variation, which are difficult to measure.
Fluorescence-inherent sources of error can be easier certified, however only, if the necessary controls and standards
are included in the experiment and if the scanners are
properly characterized. Fluorescence-based uncertainties
are related to fluorophore labeling of the target [125,126]
and to problems linked to measurements of fluorescence
intensities, i.e. the sensitivity of the label’s spectroscopic
properties to chromophore microenvironment [127,128]
and to dye–dye interactions [129] as well as to instrumentspecific effects. To overcome limitations in the former,
e.g. different incorporation rates of the dyes or dye instability, the so-called dye swap or fluor-reverse is used
with both targets being labeled with two dyes in sepa-
Resch-Genger et al.
rate reactions and hybridized to two identical microarrays with identical probe arrangements. In addition to the
instrument-related effects already discussed, microarray
scanner-specific uncertainties are linked to the determination of the background signal, the method used for dye
normalization, crosstalk between channels, and variations
in the alignment of the optical scanner [130]. For example,
for confocal reading, which is used to increase the sensitivity by rejecting light coming from other planes than
that of the biological signals of interest, there is a strong
need for accuracy of the focus and tilt adjustment of the
biochip. To illustrate the influence of such instrumentspecific effects on the recorded fluorescence intensities, a
calibration slide from Clondiag [131] was measured with
three common microarray readers with routinely used instrument parameters, e.g. representative settings of the
PMT voltage for the green and red channels, respectively.
The accordingly obtained results are depicted in Fig. 12.
The differences in fluorescence intensity measured with
all the readers for the two detection channels are attributed
Fig. 12. Comparison of three commercially available microarray readers using the calibration slide FluorIS from Clondiag, Germany. The
intensities of three areas differing in intensity (Int 1, Int 2, and Int 3) are
depicted. The measured intensities represent mean values of 100 single
spots per spot area. The slide was measured at the excitation wavelengths
typically used for Cy3 (light-gray bars) and Cy5 (dark-grey bars) at standard laboratory PMT settings. The obtained scan images were analyzed
using the IconoClust version 2.2 software package (Clondiag, Germany).
Quality Assurance in Fluorometry and Suited Fluorescence Standards
to the different wavelength-dependent spectral responsivities of the respective detection systems, see e.g. also Fig. 1.
Typically, such detection systems are less sensitive in the
long wavelength region of ca. 675 nm as compared to a detection wavelength of ca. 565 nm. The obvious deviations
between the three data sets clearly demonstrate the need
for a general calibration tool to improve the reliability of
microarray data from the detection side.
To make microarray data world-wide comparable,
there is a strong need for instrument standards, application of biological reference materials, and standardized
procedures/protocols for the performance and evaluation
of such experiments. For the reduction of biology- and
data evaluation-related uncertainties, the microarray research community has discussed since a long time the
introduction of quality controls (QCs) and SOPs as well
as biological reference materials such as for instance standards based on genomic DNA, a reference RNA pool or
a reference oligonucleotide for measuring more precisely
absolute or indirect gene expression values [132,133]. For
instance, only recently, two RNA reference materials have
been defined to improve the standardization of gene expression experiments [134]. Furthermore, working groups
have been founded to develop and share common standards and guidelines for gene expression analysis in clinical trials and to cover standards for reference materials and
quality control, SOPs as well as standard bioinformatic approaches for data analysis and storage [135–138]. Several
guidelines for the performance of microarray experiments
have been proposed by the Microarray Gene Expression
Data (MGED) Society, including the Minimal Information About a Microarray Experiment (MIAME) description and annotation [139] and the MAGE-ML mark-up
language [140]. For storage of the microarray data, several
databases have been established following the MIAME
guidelines [141,142].
To improve the reliability of microarray experiments
from the detection side, at present, there exist only very
few standards. These are intensity standards designed e.g.
for control of the performance of array scanners, channel
crosstalk, and effect of PMT voltage as well as for the
comparison of the performance of different instruments
[143]. One of them offers some means for control of the
spatial resolution [131]. This standard, however, that was
measured by us, see Fig. 12, suffers from photodegradation, lack of reported absorption and emission spectra, and
a comparatively small variation in fluorescence intensity,
see for instance [131] and Fig. 12, which does not allow
for the determination of the linearity of the detection system(s). Moreover, some microarray reader manufacturer
have integrated a self-test using an internal light source,
which runs at the start of the instruments and before every
355
scan (e.g. Fuji Photo Film Co.) to check PMT settings
whereas others (e.g. Axon Instruments Inc.) provide specific calibration slides for an evaluation of the instrument
performance.
To improve this situation, better-suited calibration
slides or calibration spots on every microarray are desired to help adjust the scanner settings and to determine the instrument’s image/spatial resolution as well
as its day-to-day performance and long-term stability—
ideally in combination with guidelines for their use. Furthermore, tested procedures in combination with recommended materials/systems are needed for the determination of the range of linearity of the array scanners, their
dynamic range and sensitivity. The final goal is here to
improve the comparability of data generated by different
instruments/laboratories and with different labels. Easyto-use fluorescence standards that meet these requirements should for instance either provide a stable output
at application-relevant wavelengths and photon fluxes or
need to be easily reproduced, should have minimum local and global non-uniformities in emission and should
generate signals comparable to those observed for typical
samples. With this respect, also the influence of spectral
deviations between the fluorescence spectra of the standard and common labels needs to be examined. This renders the availability of absorption and corrected emission
spectra of both the standard and the labels for applicationrelevant conditions, i.e. microenvironment(s), important.
For testing of the instrument’s spatial resolution, which
should be in the 5 µm range, uniformly patterned standards are desired. Such standards must be applicable for
the majority of typically used microarray readers and labels, e.g. suited for typical excitation wavelengths and
emission wavelength intervals as well as scanner formats.
Moreover, they should have been tested with a broad variety of instruments in a Round Robin test. For the development of such purpose-fit fluorescence standards, input
not only from instrument manufacturers but also from
the community of users of the microarray technology is
required. Additionally, future technological advances of
microarray readers should be considered that will most
likely provide an enhanced spatial resolution, precision,
and sensitivity as well as a higher dynamic range. Furthermore, it needs to be decided on within the community of
users of the microarray technology and instrument manufacturers, whether stand-alone fluorescence standards like
calibration slides or the integration of fluorescence standards into slides used for the performance of microarray
experiments are to be favored. The former is for instance
sufficient to characterize the day-to-day and long-term
instrument performance, whereas the latter approach—
though more costly—may be tempting for an improved
356
and better feasible comparability of the recorded fluorescence signals.
In Vivo Fluorescence Imaging/Spectroscopy
Generally, imaging of animals, human skin, and inner cavities within the human body, e.g. bladder or colon,
with endoscopic techniques is a powerful tool to detect
modifications of tissue, possibly in advance of malignant
alterations. Apart from reflection measurements, application of fluorescence imaging in medical diagnostics has attracted increasing interest over the past decade [144,145].
Since tissue is a highly scattering medium, the excitation light as well as the light emitted by the fluorophore
used as contrast agent are diffracted many times and
accordingly, the corresponding optical path-lengths are
significantly increased. Contributions from light emitted
from different depths within the tissue result in a diffuse image and hence, the technique is often termed diffuse fluorescence imaging. Optical properties of tissue
strongly depend on excitation and emission wavelengths.
Accordingly, wavelength selection influences the penetration depth and, hence, the intensity of the recorded fluorescence signal. In particular, the maximum penetration
depth and minimum absorption are reached when excitation and emission wavelengths are chosen that fall within
the so-called diagnostic and therapeutic window between
700 and 900 nm. Hence, this spectral region is used for
the detection of deeper lying lesions. For the detection
of superficial lesions, the wavelength region between 400
and 700 nm is employed as well.
In vivo fluorescence imaging allows for instance to
study the dynamics of contrast agents or specific intrinsic
fluorophores, which serve as reporters for diseased tissue
[146]. In addition, fluorescence measurements based on
specific molecular probes for target detection, i.e. molecular imaging, are used in medicine and drug design [147–
152]. Fluorescence detection typically includes measurements of the distribution of fluorescence intensities (integral, spectrally resolved, time-gated) of the corresponding
reporter molecules as well as fluorescence lifetime and occasionally fluorescence polarization analysis. The former
is strongly influenced by the absorption of the respective tissue, its light scattering properties, and the location/depth of the fluorophores in the tissue. In addition,
apart from the fluorescence of the probe molecule, background fluorescence of tissue caused by intrinsic chromophores, i.e. autofluorescence, also contributes to the
fluorescence signal. Furthermore, for fluorescence originating from deeper layers within the tissue, the covering
tissue acts as wavelength-dependent attenuator for the excitation light as well as for the emitted light. Also reflec-
Resch-Genger et al.
tion losses, which might depend on the polarization of
light, occur at boundary layers (e.g. mucosa) because of
the inhomogeneity of tissue. Accordingly, many influencing quantities have to be known to derive concentrations
of fluorophores and subsequently the concentration of target molecules from the observed fluorescence signals.
At present, this intention seems to be unrealistic. However, standardization of measurement procedures to obtain relative concentrations at higher precision is required
to ensure the comparability of measurements performed
with different setups and/or under different experimental
conditions in vivo. The application of suitable SOPs and
reference materials would facilitate the interpretation of
fluorescence images. In particular, the differentiation between a lesion and normal tissue might be improved in
this way.
Imaging of laser-induced fluorescence in tissue can
be performed in reflection [153] or transmission geometry
[154]. In both cases, a reference material with known optical, i.e. absorption, fluorescence, and scattering properties
can be measured simultaneously or subsequently using the
same geometry. A suited non-certified reference material
developed by the PTB for this purpose consists of a cyanine dye and glass spheres as scattering centers embedded
into a polymer matrix [155]. We used this reference material to derive the results shown in Fig. 13, which depicts
the measured integral fluorescence intensities from a rat
joint and fluorescence intensities normalized to this standard detected simultaneously. The contrast agent, a nonspecific cyanine dye [151,152,156] injected intravenously
prior to the experiment, was excited at 740 nm. Apart
from the large penetration depth, for the chosen excitation wavelength, the background fluorescence is very
Fig. 13. Comparison of measured (squares) and normalized (line) fluorescence intensities of a rat joint after application of the contrast agent
(non-specific cyanine dye, dose of 1 µmol/kg body weight). The normalization was done with respect to a PTB-developed reference material
simultaneously recorded. The contrast agent was excited at 740 nm.
Quality Assurance in Fluorometry and Suited Fluorescence Standards
low compared to the fluorescence intensity of the contrast
agent, see Fig. 13, observation times shorter than 15 s.
The background fluorescence from the tissue is obtained
by imaging the animal before application of the contrast
agent. As follows from Fig. 13, the integral fluorescence
intensity, plotted versus time after intravenous injection,
strongly increases between 15 and 25 s and reaches a saturation value after about 90 s. Furthermore, the fluctuations in the bolus kinetic are considerably reduced on
application of this normalization procedure as is evident
from the reduced signal variation. The remaining variation of the fluorescence signal is associated with the
heartbeat and respiration of the animal, respectively. This
referencing or normalization procedure allows to compare
fluorescence intensities from different animals provided
the normalization is performed with the same reference
material.
Fluorescence imaging and fluorescence spectroscopy can be used to extend the potential of endoscopybased techniques frequently used for clinical applications.
As an example, in Fig. 14, an endoscopic image of the
colon mucosa of a patient with pancolitis is shown. Aside
from the typical reflection image routinely observed, a
bright spot from the excitation light coupled through an
optical fiber and the fiber tip can be seen. To obtain
laser-induced fluorescence, aminolevulinic acid (ALA) is
sprayed into the colon shortly before endoscopy. ALA is
a poorly fluorescing precursor of the fluorophore proto-
Fig. 14. Fluorescence-guided endoscopy of the colon of a patient with
pancolitis. The laser light delivered by a fiber and the fiber tip can be
recognized in the lower central part of the image. This investigation
has been performed at the 4th Medical Department, Charit´e, Humboldt
University, Berlin, Germany by Dr. G. Schachschal.
357
Fig. 15. Prompt and delayed fluorescence spectrum of a dysplasia in
Barrett’s oesophagus. The background fluorescence is efficiently suppressed in the time delayed spectrum of PpIX. The arrows indicate the
wavelengths and amplitudes used for normalization.
porphyrin IX (PpIX), the formation of which is enhanced
in diseased tissue, e.g. in low- and high-grade dysplasia.
To monitor regions with enhanced PpIX concentration
on-line together with the conventional endoscopic image,
pulsed laser light for the excitation of the fluorophore and
a monochromator coupled to an intensified photodiode
array for the detection of the emitted light are used. For
excitation and observation of laser-induced fluorescence
an optical fiber is inserted into the working channel of
an endoscope. This technique enables the measurement
of spectrally resolved fluorescence data at a single point
[157,158] and can be applied for fluorescence-guided
biopsy. The fluorescence spectra from the blue spot in
Fig. 14 taken at two different delay times relative to the
exciting laser pulse are illustrated in Fig. 15. As is evident
from a comparison of the prompt and delayed emission
spectra, application of this time-gated technique results in
an effective suppression of tissue autofluorescence. Since
geometrical factors remain unchanged, analysis of both
spectra can be exploited to eliminate geometrical effects,
which otherwise hamper the interpretation and quantification of the measurement. To this end [158], we introduce
the Pp IX-specific quantity
In (λ, 20 ns)
sp
In (λ, 20 ns) = In (599 nm, 20 ns)
In (599 nm, 20 ns)
In (λ, 0 ns)
(1)
−
In (599 nm, 0 ns)
In Eq. (1), the prompt I(λ, 0 ns) and delayed I(λ,
20 ns) fluorescence spectra, see Fig. 15 are normalized to
the maximum intensity I(635 nm, 0 ns) of the prompt fluorescence spectrum. The resulting normalized quantities
are indicated by the lower index n. The normalization to
358
I(635 nm, 0 ns) is chosen with respect to the maximum
emission of PpIX and the division by the spectral amplitude at λ = 599 nm is performed because non-specific
fluorescence and PpIX fluorescence spectra do not overlap in this region. In this way, the contribution from nonspecific autofluorescence can be subtracted, provided that
the optical properties of diseased and normal tissue are
similar. This assumption is generally justified for the detection of early stages of cancerous tissue. On the basis of
this normalization procedure, a significant discrimination
between diseased and non-diseased tissue is possible as
has been shown for example by Ortner et al. for dysplasias
in Barrett’s oesophagus [157].
For fluorescence imaging in vivo, for spectral correction as well as for the determination of the linearity of
the detection system, its dynamic range, and its sensitivity, solutions of the corresponding contrast agents or other
spectrally suited dyes could be used. However, to improve
the comparability and reliability of in vivo applications of
fluorescence imaging, on our opinion, the use of reference
procedures including mathematical analysis and properly
chosen fluorescence standards seems to be the method of
choice, as has been demonstrated exemplary for molecular imaging experiments with small animals and fluorescence endoscopy of human colon. Reference materials
suited for signal quantification must also mimic the scattering properties of the investigated tissue. To this end, for
the comparability of different in vivo applications, a set
of well-characterized and preferably certified standards
of liquid or solid nature is desired that covers the spectral
range from 400 up to 900 nm.
CONCLUSION AND OUTLOOK
As has been illustrated for selected fluorescence techniques, there is still a considerable need to improve the reliability and the comparability of fluorescence data as well
as the comparability of instrument characterization and
performance validation. The latter forms the basis for standardization. To achieve this and to improve quality assurance in fluorometry, well-characterized instrument-type
and application-specific fluorescence standards for the
UV/Vis/NIR spectral region are mandatory, that closely
match the optical properties of measured samples. In
this context, close resemblance stands for comparable absorption and emission features, emission intensities and
characteristics, scattering and measurement geometry. In
addition to (certified) reference materials, internationally
accepted SOPs for the characterization and validation of
common types of fluorescence instruments are needed as
well as guidelines for the performance of fluorescence
Resch-Genger et al.
measurements that are relevant for a broad community of
users of fluorescence techniques. The intention of the latter should be also to reduce sample-related and handlingrelated uncertainties in fluorescence analysis. On our opinion, guidelines—preferably in combination with suited
materials or standards—are for instance required for the
determination of the range of linearity, dynamic range, and
sensitivity of fluorescence instruments as well as for the
performance of quantitative fluorescence measurements
and for the determination of fluorescence quantum yields.
The availability of suited certified standards and guidelines is of special importance for fluorescence measurements in medical diagnosis and clinical applications of
fluorometry, where standardization of instrument characterization and performance of measurements is eventually
requested.
Novel fluorescence standards for the establishment
of an improved quality assurance in fluorometry on
a broad level should be not only designed and characterized for use under application-relevant conditions,
easy-to-operate, and commercially available, but also
characterized with respect to all the application-relevant
spectroscopic and analytical properties. Furthermore, the
(wavelength-dependent) uncertainty of the calibrationrelevant fluorometric quantity/quantities needs to be provided including method of determination. To guarantee
the standard’s reliability, such reference materials should
be preferably certified. The rational design and testing of
such fluorescence standards with improved properties and
the development of the aforementioned guidelines, however, requires a close contact between National Metrological Institutes (NMIs), standardization bodies, regulatory
agencies, scientific associations, manufacturers of instrumentation, users of fluorescence techniques as well as
manufacturers of fluorescent labels, reagents and other
materials for fluorescence analysis. Within this context,
also the necessary level of traceability in fluorometry has
to be defined and recent developments in instrumentation as well as in fluorescent dyes need to be taken into
account. Particular attention should be focused also on
procedures for the evaluation of software used for data
analysis, at least for more advanced fluorescence methods
like time-resolved fluorometry, fluorescence imaging, and
fluorescence-based microarray techniques.
ACKNOWLEDGMENTS
Financial support from the German Ministry of Economics and Labor (BMWA) is gratefully acknowledged
(BMWA VI A2-17/03). We thank Mrs. A. Hoffmann and
Mrs. M. Spieles for technical assistance, Dr. D. Pfeifer for
Quality Assurance in Fluorometry and Suited Fluorescence Standards
stimulating discussions and Dr. J. Enderlein for carefully
reading the section on fluorescence microscopy.
REFERENCES
1. J. R. Lakowicz (Ed.) (1999). Principles of Fluorescence Spectroscopy, 2nd edn., Kluwer Academic/Plenum Press, New York.
2. J. R. Lakowicz (Ed.) (1992–2004). Topics in Fluorescence Spectroscopy Series, Vols. 1–8, Plenum Press, New York.
3. O. S. Wolfbeis (Series Ed.) (2001–2004). Springer Series on Fluorescence, Methods and Applications, Vols. 1–3, Springer, Berlin.
4. S. G. Schulman (Ed.) (1985–1993). Molecular Luminescence Spectroscopy Parts, Vols. 1–3, Wiley Interscience, New York.
5. W. T. Mason (1999). Fluorescent and Luminescent Probes for Biological Activity, 2nd edn., Academic Press, San Diego.
6. A. J. Pope, U. M. Haupts, and K. J. Moore (1999). Homogenous
fluorescence readouts for miniaturized high-throughput screening:
Theory and practice. Drug Discov. Today 4(8), 350–362.
7. E. Zubritsky (1999). Microplate readers reach critical mass. Anal.
Chem. News Features 71, 39A–43A.
8. C. T. Wittwer, K. M. Ririe, R. V. Andrew, D. A. David, R. A.
Gundry, and U. J. Balis (1997). The light cycler: A microvolume
multisample fluorimeter with rapid temperature control. Biotechniques 22, 176–181.
9. Supplement (1998). Fluoreszenzspektroskopie. Nachr. Chem. Tech.
Lab. 46, S121–S132.
10. Analytical Methods Committee (1998). Evaluation of analytical instrumentation. Part XI Instrumentation for molecular fluorescence
spectrometry. Analyst 123, 1649–1656.
11. J. W. Eastman (1966). Standardization of fluorescence spectra and
the calibration of spectrofluorimeters. Appl. Optics 5(7), 1125–
1132.
12. W. Galbraith, K. W. Ryan, N. Gliksman, D. L. Taylor, and A. S.
Waggoner (1989). Multiple spectral parameter imaging in quantitative fluorescence microscopy. I: Quantitation of bead standards.
Comput. Med. Imag. Graph. 13, 47–60.
13. A. K. Gaigalas, L. Li, O. Henderson, R. Vogt, J. Barr, G. Marti, J.
Weaver, and A. Schwartz (2001). The development of fluorescence
intensity standards. J. Res. Natl. Inst. Stand. Technol. 106(2), 381–
389.
14. C. A. Parker (1968). Photoluminescence of Solutions, Elsevier,
Amsterdam.
15. J. N. Miller (1981). Standards in Fluorescence Spectrometry, Ultraviolett Spectrometry Group, London.
16. D. F. Eaton (1988). Reference materials for fluorescence measurement. Pure Appl. Chem. 60, 1107–1114.
17. R. A. Velapoldi and M. S. Epstein (1989). In M. C. Goldberg
(Ed.), Luminescence Applications in Biological, Chemical, Environmental and Hydrological Sciences, ACS Symposium Series,
Vol. 383, American Chemical Society, Washington, DC, pp. 98–
126.
18. W. D. Niles and F. S. Cohen (1995). Radiometric calibration
of a video fluorescence microscope for the quantitative imaging of resonance energy transfer. Rev. Sci. Instrum. 66, 3527–
3536.
19. EN ISO/IEC 17025; GLP/GMP, GLP: Good laboratory praxis;
GMP: Good manufacturing praxis.
20. Burgess and D. G. Jones (1995). Spectrophotometry, Luminescence
and Colour: Science and Compliance, Elsevier, Amsterdam.
21. O. D. D. Soares and J. L. C. Costa (1999). Spectrophotometers
intercomparison for spectrocolorimetric scale harmonization. Rev.
Sci. Instrum. 70(12), 4471–4481.
22. J. C. Travis, J. C. Zwinkels, F. Mercader, A. Ruiz, E. A. Early,
M. V. Smith, M. Noel, M. Maley, G. W. Kramer, K. L. Eckerle,
and D. L. Duewer (2002). An international evaluation of holmium
oxide solution reference materials for wavelength calibration in
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
359
molecular absorption spectrophotometry. Anal. Chem. 74, 3408–
3415.
ASTM E 169-87 (reapproved 2003). Standard Practices for general
Techniques of Ultraviolet-Visible Quantitative Analysis, and herein
referenced ASTM standards.
CIE-Publ. 15.2, Colorimetry, 2nd edn., 1986.
K. D. Mielenz (1987). In C. Burgess and K. D. Mielenz (Eds.),
Advances in Standards and Methodology in Spectrophotometry,
Elsevier, Amsterdam, pp. 49–62.
D. C. Rich and D. Martin (1999). Improved model for improving
the inter-instrument agreement of spectrocolorimeters. Anal. Chim.
Acta 380, 263–276.
K. Witt (2002/2003). Haben wir die UV/vis-Spektrometerie lumineszierender Materialien im Griff? Die Farbe 46(1/2), 31–51.
ASTM E 388-72 (reapproved 2003). Spectral bandwidth and wavelength accuracy of fluorescence spectrometers.
ASTM E 578-83 (reapproved 2003). Linearity of fluorescence measuring system.
ASTM E 579-84 (reapproved 2003). Limit of detection of fluorescence of quinine sulfate.
A. Schwartz and E. Fernandez-Repollet (1993). Development of
clinical standards for flow cytometry. Ann. N. Y. Acad. Sci. 677,
28–39.
A. Schwartz, M. Mendez, G. Santiago, L. Diaz, and E. FernandezRepollet (1997). Applications of common quantitative fluorescent
standards to multiple platforms: Comparison of commercial fluorescent calibration standards used in quantitative flow cytometry.
Clin. Immunol. 17(1), 14–18.
A. Schwartz, G. E. Marti, J. W. Gratama, and E. Fernandez-Repollet
(1998). Standardizing flow cytometry: A classification system of
fluorescence standards used for flow cytometry. Cytometry 33, 106–
114.
J. W. Gratama, J. L. D’Hautcourt, F. Mandy, G. Rothe, D. Barnett,
G. Janossy, S. Papa, G. Schmitz, and R. Lenkei (1998). Flow cytometric quantitation of immunofluorescence intensity: Problems
and perspectives. Cytometry 33, 166–178.
R. A. Velapoldi (1987). Liquid standards in fluorescence spectrometry. In C. Burgess and K. D. Mielenz (Eds.), Advances in Standards
and Methodology in Spectrophotometry, Elsevier, Amsterdam,
pp. 175–193.
I. Billard, E. Ansoborlo, K. Afferson, S. Arpigny, M. E. Azenha,
D. Brich, P. Bros, H. D. Burrows, G. Choppin, L. Couston,
V. Dubois, T. Fangh¨anel, G. Geipel, S. Hubert, J. I. Kim, T.
Kimura, R. Klenze, A. Kronenberg, M. Kumke, G. Lagarde, G.
Llamarque, S. Lis, C. Madic, G. Meinrath, C. Moulin, R. Nagaishi,
D. Parker, G. Plancque, F. Scherbaum, E. Simoni, S. Sinkov, and
C. Viallesoubranne (2003). Appl. Spectrsc. 57(8), 1027–1038.
For clear definition of the metrological hierarchy of reference materials, see International Vocabulary of Basics and General Terms
in Metrology (1994), 2nd edn., Beuth Verlag GmbH.
R. A. Velapoldi and K. D. Mielenz (1980). A fluorescence standard
reference material: Quinine sulfate dihydrate, NBS Spec. Publ. 260264, PB 80132046, Springfield, VA.
A. Schwartz, L. Wang, E. Early, A. Gaigalas, Y.-Z. Zhang, G. E.
Marti, and R. F. Vogt (2002). Quantitating fluorescence intensity
from fluorophore: The definition of MESF assignment. J. Res. Natl.
Inst. Stand. Technol. 107(1), 83–91.
U. Resch-Genger, D. Pfeifer, C. Monte, W. Pilz, A. Hoffmann, M.
Spieles, J. Hollandt, R. D. Taubert, B. Sch¨onenberger, P. Nording
(in press), Traceability of fluorometry. Part II: Spectral fluorescence
standards. J. Fluoresc. 15(3), 325.
J. W. Verhoeven (1996). Molecular terms used in photochemistry
(recommendations 1996). Pure Appl. Chem. 68(12), 2223–2286.
J. Hollandt, R. D. Taubert, J. Seidel, A. Gugg-Helminger, U. ReschGenger, D. Pfeifer, C. Monte, and W. Pilz (in press), Traceability
of fluorometry. Part I: Physical standards. J. Fluoresc. 15(3), 311.
E. D. Cehelnik, K. D. Mielenz, and R. A. Velapoldi (1975). Polarization effects on fluorescence measurements. J. Res. Natl. Bur.
Stand. A 79A(1), 1–15.
360
44. U. Resch-Genger, K. Hoffmann, and A. Engel, manuscript in
preparation.
45. T. Erdogan, A. Pradhan, and V. Mizrahi (2003). Optical filters
impact fluorescence fidelity. Biophotonics Int. 10(10), 38–43.
46. J.W. Hofstraat and M. J. Latuhihin (1994). Correction of fluorescence spectra. Appl. Spectrosc. 48(4), 436–446.
47. J. A. Gardecki and M. Maroncelli (1998). Set of secondary emission
standards for calibration of the spectral responsivity in emission
spectroscopy. Appl. Spectrosc. 52(9), 1179–1189.
48. R. J. Kovach and W. M. Peterson (1994). The measurement of
sensitivity in fluorescence spectroscopy. Am. Lab. 32G–32K.
49. R. B. Thompson, I. Gryczynski, and J. Malicka (2002). Fluorescence polarization standards for high-throughput screening and
imaging. Biotechniques 32(1), 34–41.
50. See for instance Molecular Probes, Starna GmbH, Matech Precision Dynamics Coorp., Labsphere Inc., Fluka GmbH, LambdaChem GmbH, and SUMITA Optical Glass Inc. as well as NIST
(RM 8640, SRM 936a, and 1932).
51. R. F. Chen (1972). Measurement of absolute values in biochemical
fluorescence spectroscopy. J. Res. Nat. Bur. Stand. A 76(6), 593–
606.
52. R. B. Thompson, I. Gryczynski, and J. Malicka (2002). Fluorescence polarization standards for high-throughput screening and
imaging. Biotechniques 32(1), 34–41.
53. I. T. Lifshotz and M. L. Meilman (1989). Standard sample for
calibrating wavelength scales of spectral fluorimeters. Sov. J. Opt.
Technol. 55(8), 487–489.
54. S. A. Wise, L. C. Sander, and W. E. May (1993). Determination
of polycyclic aromatic hydrocarbons by liquid chromatography.
J. Chromatogr. 642, 329–349.
55. SRM 1647b, NIST.
56. A. Schwartz, A. K. Gaigalas, L. Wang, G. E. Marti, R. F. Vogt, and
E. Fernandez-Repollet (2004). Formalization of the MESF unit of
fluorescence intensity. Cytometry 57B(1), 1–6.
57. RM 8640, NIST.
58. P. Froehlich (1989). Under the sensitivity specification for a fluorescence spectrophotometer. Int. Lab. 42–44.
59. R. J. Kovach and W. M. Peterson (1994). The measurement of
mensitivity in fluorescence spectroscopy. Am. Lab. 32G–32K.
60. ISO (1993). Guide to the expression of uncertainty in measurement.
61. J. N. Demas (1982). In K. D. Mielenz (Ed.), Optical Radiation
Measurements, Vol. 3, Academic Press, New York, p. 195.
62. W. Geffken (1962). The molar absorption of different ions in
glasses. Glastechn. Berichte 35, 27–35.
63. M. Mizuguchi, H. Hosono, and H. Kawazone (1999). Timeresolved photoluminescence for diagnostic of photoluminescence
to ArF excimer laser damage to CaF2 single crystals. J. Opt. Soc.
Am. 7(16), 1153–1159.
64. P. de Rose (2003). Bioanalytical and biomedical applications of
fluorescence techniques: Instrument characterization and validation, traceability, and need for reference materials, in Fluorescence
Workshop, BERM-9, Berlin.
65. D. Ehrt, P. Ebeling, U. Natura, U. Kohlberg, K. Naumann, and S.
Ritter (2001). Redox equilibria and ultraviolet radiation induced
defects in glasses. Int. Cong. Glass 1, 84–96.
66. J. W. Chan, T. Huster, J. S. Hayden, S. H. Risbud, and D. M. Krol
(2002). J. Am. Ceram. Soc. 85(5), 1037–1040.
67. A. Engel, K. Knapp, B. Speit, G. Wehrhan, and E. M¨orsen (2001).
High quality CaF2 used for 157 nm micro lithography. Fab. Tech.
14, 177–184.
68. C. M¨uhlig, W. Triebel, G. T¨opfer, and A. Jordanov (2003). CaF2
for ArF lithography—Characterisation by in-situ and LIF measurements, CHOCLAB II final report—Optics Characterization,
pp. 257–267.
69. A. Engel, W. Triebel, C. M¨uhlig, J. Alkemper, A. Kr¨amer, J.
Kandler, K. Knapp, and E. M¨orsen (2000). Visualization of laser
damage in 157 nm material CaF2 and BaF2 , 1st 157 nm Symposium,
Dana Point CA, pp. 391–398.
Resch-Genger et al.
70. A. Engel, R. Haspel, and V. Rupertus (2003). Advanced industrial fluorescence metrology used for qualification of high-quality
optical materials. SPIE Proc. 5118-5120, 182–189.
71. R. A. Velapoldi (1971). Fluorescence. Nat. Bur. Stand. Tech. Note
584, 53–83.
72. H. Pick (1972). Structure of trapped electron and trapped hole
centers in alkali halides. In J. Abeles (Ed.), Optical Properties of
Solids, North Holland, pp. 653–668.
73. D. W. Pack, W. J. Manthey, and D. S. Mc Clure (1989). Production
of color centers with ionizing irradiation in alkali halides. Phys.
Rev. B 40(14), 9930–9935.
74. M. Letz, A. Engel, L. Parthier, U. Natura, and K. Knapp
(2004). CaF2 for DUV lens fabrication: Basic material properties
and dynamic light-matter interaction. SPIE Proc. 5377, Optical
Microlithography XVII, 1797–1804.
75. Japanese Optical Glass Industrial Standards, JOGIS, 03-1975.
76. W. Gellermann (1989). J. Chem. Solids 52, 249–254.
77. W. Goehde, U. Cassens, L. G. Lehman, Y. Traore, W. Goehde
jun., P. Berkes, C. Westerberg, and B. Greve (2003). Individual
patient-dependent influence of erythrocyte lysing procedures on
flow-cytometric analysis of leukocyte subpopulations. Transfusion
Med. Hemother. 30, 165–170.
78. J.-C. Strohmeyer, C. Blume, C. Meisel, W.-D. Doecke, M. Hummel,
C. Hoeflich, K. Thiele, A. Unbehaun, R. Hetzer, and H.-D. Volk
(2003). Standardized immune monitoring for the prediction on
infections after cardiopulmonary bypass surgery in risk patients.
Cytometry 53B, 54–62.
79. G. Monneret, N. Elmenkouri, J. Bohe, A. L. Debard, A. C.
Gutowski, J. Bienvenu, and A. Lepape (2002). Analytical requirements for measuring monocytic human lymphocyte antigen DR
by flow cytometry: Application to the monitoring of patients with
septic shock. Clin. Chem. 48, 1589–1592.
80. S. B. Iyer, M. J. E. Bishop, B. Abrams, V. C. Maino, A. J.
Ward, T. P. Christion, and K. A. Davis, QuantiBRITETM : A new
standard for fluorescence quantification, www.bdbiosciences.com/
immunocytometry systems (see download literature, White Papers,
QuantiBRITETM White Paper).
81. Y. Gerena-L´opez, J. Nolan, L. Wang, A. Gaigalas, A. Schwartz, and
E. Fern´andez-Repollet (2004). Quantification of EGFP expression
on Molt-4 T cells using calibration standards. Cytometry 60A, 21–
28.
82. J. B. Pawley (Ed.) (1995). Handbook of Biological Confocal Microscopy, 2nd edn., Kluwer Academic Publishers, New York.
83. S. Inoue (Ed.) (1986). Video Microscopy, Plenum Publishers, New
York.
84. D. B. Murphy (Ed.) (2001). Fundamentals of Light Microscopy
and Electronic Imaging, Wiley-Liss, New York.
85. X. F. Wang, A. Periasamy, B. Herman, and D. M. Coleman
(1992). Fluorescence lifetime imaging microscopy (FLIM): Instrumentation and applications. Crit. Rev. Anal. Chem. 23(5), 369–
395.
86. Special issue on the use of ion-sensitive fluorophores for making accurate intracellular ion measurements at high spatial and/or
temporal resolution (1990). Cell Calcium, February/March.
87. M. Andreeff and D. Pinkel (Ed.) (1999). Introduction to Fluorescence in Situ Hybridization: Principals and Clinical Applications,
Wiley-Liss, New York.
88. V. E. Centonze, A. Takahashi, E. Casanova, and B. Herman (2000).
Quantitative fluorescence microscopy. J. Histotechnol. 23(3), 229–
234.
89. W. D. Niles and F. C. Cohen (1995). Radiometric calibration of
a video fluorescence microscope for quantitative imaging of resonance energy transfer. Rev. Sci. Instrum. 66(6), 3527–3536.
90. R. Nitschke (2004). Standardization and quantification in microscopy, Workshop AK PhotonicNet, Wetzlar.
91. The point-spread function is determined by the product of the
excitation intensity distribution and the light collection efficiency
function.
Quality Assurance in Fluorometry and Suited Fluorescence Standards
92. J. S. Ploem (1970). Standards for fluorescence microscopy, in E. J.
Holborow (Ed.), Standard for Immunofluorescence Symposium,
Blackwell Scientific Publications, Oxford, pp. 137–153.
93. W. Galbraith, K. W. Ryan, N. Gliksman, D. Lansing Taylor, and
A. S. Waggoner (1989). Multiple spectral parameter imaging in
quantitative fluorescence microscopy. I: Quantitation of bead standards. Comp. Med. Imag. Graphics 13(1), 47–60.
94. J. M. Lerner and R. M. Zucker (2004). Calibration and validation
of confocal spectral imaging systems. Cytometry 62A, 8–34.
95. M. Sernetz and A. Thaer (1970). A capillary fluorescence standard
for microfluorometry. J. Microscopy 91(1), 43–52.
96. F. W. D. Rost (1991). Quantitative Fluorescence Microscopy,
Cambridge University Press, Cambridge, p. 236.
97. D. S. Kaplan and G. L. Picciolo (1989). Characterization of instrumentation and calibrators for quantitative microfluorometry for
immunofluorescence tests. J. Clin. Microbiol. 27, 442–447.
98. A. P. M. Jongsma, W. Hijmans, and J. S. Ploem (1971). Quantitative
immunofluorescence. Histochemie 25, 329–343.
99. R. A. Velapoldi, J. C. Travis, W. A. Cassatt, and W. T. Yap (1975).
Inorganic ion-doped glass fibres as microspectrofluorimetric standards. J. Microsc. 103(3), 293–303.
100. R. P. Haugland (Ed.) (2002). Handbook of Fluorescent Probes and
Research Products, 9th edn., Molecular Probes, Section 24.1.
101. J. J. Haaijman and J. P. R. van Dalen (1974). Quantification in
immunofluorescence microscopy: A new standard for fluorescein
and rhodamine emission measurement. J. Immunol. Methods 5,
359–374.
102. S. J. Lockett, K. Jacobson, and B. Herman (1992). Quantitative
precision of an automated fluorescence-based image cytometer.
Anal. Quant. Cytol. Histol. 14, 187–202.
103. M. A. Model and J. K. Burkhardt (2001). A standard for calibration
and shading correction of a fluorescence microscope. Cytometry 44,
309–316.
104. J. E. Sisken (1989). Fluorescent standards. Methods Cell Biol. 30,
113–126.
105. S. G. Turney, S. M. Culican, and J. W. Lichtman (1996). A quantitative fluorescence-imaging technique for studying acetylcholine
receptor turnover at neuromuscular junctions in living animals.
J. Neurosci. Methods 64, 199–208.
106. P. W. Stevens and D. M. Kelso (2003). Imaging and analysis of
immobilized particle arrays. Anal. Chem. 75, 1147–1154.
107. A. C. Jones, M. Millington, J. Muhl, J. M. De Freitas, J. S. Barton,
and G. Gregory (2001). Calibration of an optical fluorescence
method for film thickness measurement. Meas. Sci. Technol. 12,
N23–N27.
108. Evident Technologies, Product Catalogue, February 2004.
109. A. P. Alivisatos (1996). Semiconductor clusters, nanocrystals, and
quantum dots. Science 271, 934–937.
110. A. Knight, J. Gaunt, T. Davidson, V. Chechik, and S. Windsor
(2004). Evaluation of the suitability of quantum dots as fluorescence standards. NPL report DQL-AS 007.
111. W. G. J. H. M. van Sark, P. L. T. M. Frederix, D. J. van den
Heuvel, H. C. Gerritsen, A. A. Bol, J. N. J. van Lingen, C. de Mello
Donega, and A. Meijerink (2001). Photoxidation and photobleaching of single CdSe/ZnS quantum dots probed by room-temperature
time resolved spectroscopy. J. Phys. Chem. B 105(10), 8281–
8284.
112. I. T. Young (1983). The use of digital image processing techniques
for the calibration of quantitative microscopes. Proc. SPIE 387,
326–335.
113. APE GmbH, Berlin; www.ape-berlin.de.
114. M. C. Pirrung (2002). How to make a DNA chip. Angew. Chem.
Int. Ed. 41(8), 1276–1289.
115. J. R. Epstein, I. Biran, and D. R. Walt (2002). Fluorescence-based
nucleic acid detection and microarrays. Anal. Chim. Acta 469, 3–
36.
116. B. Lemieux, A. Asharoni, and M. Schena (1998). Overview of
DNA chip technology. Mol. Breed. 4, 277–289.
361
117. P. Hedge, R. Qi, K. Abernathy, C. Gay, S. Dharap, R. Gaspard,
J. E. Hughes, E. Snesrud, N. Lee, and J. Quackenbush (2000). A
concise guide to cDNA microarray analysis. Biotechniques 29(3),
548–562.
118. C. B. V. Christensen (2002). Arrays in biological and chemical
analysis. Talanta 56, 289–299.
119. Biochips (Market survey) (2003). New Drugs 26–29.
120. M. Schena, D. Shalon, R. W. Davis, and P. O. Brown (1995).
Quantitative monitoring of gene expression patterns with a complementary DNA microarray. Science 270(5235), 467–470.
121. M. B. Eisen, P. T. Spellman, P. O. Brown, and D. Botstein (1998).
Cluster analysis and display of genome-wide expression patterns.
Proc. Natl. Acad. Sci. U.S.A. 95(25), 14863–14868.
122. S. Granjeaud, F. Bertucci, and B. R. Jordan (1999). Expression
profiling: DNA arrays in many guises. Bioassays 21(9), 781–790.
123. M. B. Eisen and P. O. Brown (1999). DNA arrays for analysis of
gene expression. Meth. Enzymol. 303, 179–205.
124. A. Butte (2002). The use and analysis of microarray data. Nat. Rev.
Drug Discov. 1, 951–960
125. M. Taniguchi, K. Miura, H. Iwao, and S. Yamanaka (2001). Quantitative assessment of DNA microarrays—Comparison with Northern Blot analyses. Genomics 71, 34–39.
126. M. Bartosiewicz, M. Trounstine, D. Barker, R. Jonston, and A.
Buckpitt (2000). Development of a toxicological gene array and
quantitative assessment of this technology. Arch. Biochem. Biophys.
376(1), 66–73.
127. E. A. Winzeler, M. Schena, and R. W. Davis (1999). Fluorescencebased expression monitoring using microarrays. Meth. Enzymol.
306, 3–18.
128. Z. Guo, R. A. Guilfoyle, A. J. Thiel, R. Wang, and L. M. Smith
(1994). Direct fluorescence analysis of genetic polymorphism by
hybridization with oligonucleotide arrays an glass supports. Nucleic Acids Res. 22(24), 5456–5465.
129. J. B. Randolph and A. S. Waggoner (1997). Stability, specifity
and fluorescence brightness of multiply-labeled fluorescent DNA
probes. Nucleic Acids Res. 25(14), 2923–2929.
130. F. Perraut, A. Lagrange, P. Pouteau, O. Peyssonneaux, P. Puget, G.
McGall, L. Menou, R. Gonzalez, P. Labeye, and F. Ginot (2002). A
new generation of scanners for DNA chips. Biosens. Bioelectron.
17, 803–813.
131. K. Adelheim, E. Emantraut, T. Kaiser, and J. Tuchscheerer (2002).
Smart chip for array experiment standardization. New Drugs 22–
23.
132. A. M. Dudley, J. Aach, M. A. Steffen, and G. M. Church (2002).
Measuring absolute expression with microarrays with a calibrated
reference sample and an extended signal intensity range. Proc. Natl.
Acad. Sci. U.S.A. 99(11), 7554–7559.
133. M. R. Weil, T. Macatee, and H. R. Garner (2002). Toward a universal standard: Comparing two methods for standardizing spotted
microarray data. Biotechniques 32(6), 1310–1314.
134. M. Cronin, K. Ghosh, F. Sistare, J. Quackenbush, V. Vilker, and C.
O’Connell (2004). Universal RNA Reference Materials for Gene
Expression. Clin. Chem. [meeting review].
135. Y. F. Leung and D. Cavalieri (2003). Fundamentals of cDNA microarray data analysis. Trends Genet. 19(11), 649–659.
136. T. Forster, D. Roy, and P. Ghazal (2003). Experiments using microarray technology: Limitations and standard operating procedures. J. Endocrinol. 178(2), 195–204.
137. Tumor Analysis Best Practices Working Group (2004). Expression
profiling—Best practices for data generation and interpretation in
clinical trials. Nat. Rev. Genet. 5(3), 229–237.
138. A. T. Weeraratna, J. E. Nagel, V. d. V. Mello-Coelho, and D. D. Taub
(2004). Gene expression profiling: from microarrays to medicine.
J. Clin. Immunol. 24(3), 213–224.
139. A. Brazma, P. Hingamp, J. Quackenbush, G. Sherlock, P. Spellman,
C. Stoeckert, J. Aach, W. Ansorge, C. A. Ball, H. C. Causton, T.
Gaasterland, P. Glenisson, F. C. Holstege, I. F. Kim, V. Markowitz,
J. C. Matese, H. Parkinson, A. Robinson, U. Sarkans, S. Schulze-
362
140.
141.
142.
143.
144.
145.
146.
147.
Resch-Genger et al.
Kremer, J. Stewart, R. Taylor, and J. Vilo (2001). Vingron M.
Minimum information about a microarray experiment (MIAME)toward standards for microarray data. Nat. Genet. 29(4), 365–371.
P. T. Spellman, M. Miller, J. Stewart, C. Troup, U. Sarkans,
S. Chervitz, D. Bernhart, G. Sherlock, C. Ball, M. Lepage,
M. Swiatek, W. L. Marks, J. Goncalves, S. Markel, D. Iordan,
M. Shojatalab, A. Pizarro, J. White, R. Hubley, E. Deutsch, M.
Senger, B. J. Aronow, A. Robinson, D. Bassett, C. J. Jr. Stoeckert,
and A. Brazma (2002). Design and implementation of microarray gene expression markup language (MAGE-ML). Genome Biol.
3(9), RESEARCH0046.
A. Brazma, H. Parkinson, U. Sarkans, M. Shojatalab, J. Vilo, N.
Abeygunawardena, E. Holloway, M. Kapushesky, P. Kemmeren,
G. G. Lara, A. Oezcimen, P. Rocca-Serra, and S. A. Sansone (2003).
ArrayExpress—A public repository for microarray gene expression
data at the EBI. Nucleic Acids Res. 31(1), 68–71.
J. Gollub, C. A. Ball, G. Binkley, J. Demeter, D. B. Finkelstein,
J. M. Hebert, T. Hernandez-Boussard, H. Jin, M. Kaloper, J. C.
Matese, M. Schroeder, P. O. Brown, D. Botstein, and G. Sherlock
(2003). The Stanford Microarray Database: Data access and quality
assessment tools. Nucleic Acids Res. 31(1), 94–96.
See for example, Clondiag Chip Technologies GmbH; Full Moon
Biosystems Inc.
G. A. Wagni`eres, W. M. Star, and B. C. Wilson (1998). In vivo fluorescence spectroscopy and imaging for oncological applications.
Photochem. Photobiol. 68, 603–632.
R. Richards-Kortum and E. M. Sevick-Muraca (1996). Quantitative
optical spectroscopy for tissue diagnosis. Annu. Rev. Phys. Chem.
47, 555–606.
D. J. Cruccia, F. Bevilacqua, A. J. Durkin, S. Merritt, B. J.
Tromberg, G. Gulsen, H. Yu, J. Wang, and O. Nalcioglu (2003). In
vivo quantification of optical contrast agent dynamics in rat tumors
by use of diffuse optical spectroscopy with magnetic resonance
imaging coregistration. Appl. Opt. 42, 2940–2950.
D. Y. Paithankar, A. U. Chen, B. W. Pogue, M. S. Patterson, and
E. M. Sevick-Muraca (1997). Imaging of fluorescent yield and lifetime from multiply scattered light reemitted from random media.
Appl. Opt. 36, 2260–2272.
148. R. Weissleder and U. Mahmood (2001). Mol. Imag. Radiol. 219,
316–333.
149. V. Ntziachristos, J. Ripoll, and R. Weissleder (2002). Would nearinfrared fluorescence signals propagate through large human organs
for clinical studies? Opt. Lett. 27, 333–335.
150. R. Weissleder and V. Ntziachristos (2003). Shedding light onto live
molecular targets. Nat. Med. 9, 123–128.
151. M. Rudin and R. Weissleder (2003). Molecular imaging in drug
discovery and development. Nat. Rev. Drug Discov. 2, 123–
131.
152. V. Ntziachhristos, C. Bremer, and R. Weissleder (2003). Fluorescence imaging with near-infrared light: New technological advances that enable in vivo molecular imaging. Eur. Radiol. 13,
195–208.
153. A. Becker, C. Hessenius, K. Licha, B. Ebert, U. Sukowski, W.
Semmler, B. Wiedenmann, and C. Gr¨otzinger (2001). Receptortargeted optical imaging of tumors with near-infrared fluorescent
ligands. Nat. Biotechnol. 19, 327–331.
154. E. E. Graves, J. P. Culver, J. Ripoll, R. Weissleder, and V.
Ntziachristos (2004). Singular-value analysis and optimization of
experimental parameters in fluorescence molecular tomography.
J. Opt. Soc. Am. A Opt. Image Sci. Vis. 21, 231–241.
155. U. Sukowski, F. Schubert, D. Grosenick, and H. Rinneberg (1996).
Preparation of solid phantoms with defined scattering and absorption properties for optical tomography. Phys. Med. Biol. 41, 1823–
1844.
156. K. Licha (2002). Contrast agents for optical imaging. Topics Curr.
Chem. 222, 22–29.
157. M.-A. E. J. Ortner, B. Ebert, E. Hein, K. Zumbusch, D. Nolte,
U. Sukowski, J. Weber-Eibel, B. Fleige, M. Dietel, M. Stolte,
G. Oberhuber, R. Porschen, B. Klump, H. H¨ortnagel, H. Lochs,
and H. Rinneberg (2003). Time gated fluorescence spectroscopy in
Barrett’s oesophagus. Gut 52, 28–33.
158. K. T. Moesta, B. Ebert, T. Handke, D. Nolte, C. Nowak, W. E.
Haensch, R. K. Pandey, T. J. Dougherty, H. Rinneberg, and
P. M. Schlag (2001). Protoporphyrin IX occurs naturally in
colorectal cancers and their metastases. Cancer Res. 61, 991–
999.